Dissertation_Wen.pdf

INTEGRATED RETINAL IMPLANTS

Thesis by Wen Li

In Partial Fulfillment of the Requirement for the Degree of

Doctor of Philosophy

CALIFORNIA INSTITUTE OF TECHNOLOGY

Pasadena, California

2009

(Defended October 8, 2008)

ii

© 2009

Wen Li

All Rights Reserved

iii

To my family

iv ACKNOWLEDGEMENTS

This dissertation is the product of years of efforts since I came to Caltech as a graduate student. During my time here, I have encountered a number of people who have given me tremendous help to make all this happen. I would like to thank them for their guidance, support, and friendship in this humble acknowledgment. First and foremost, I would like to express my sincerest gratitude to my advisor, Prof. Yu-Chong Tai, for providing me with the opportunity, guidance, and resources to complete my Ph.D. work. His admirable expertise, never-ending supply of new ideas and passion in research has greatly inspired me and enriched my growth as a student and a researcher. Dr. Tai is not only a great scientist, but also a respectable mentor, who taught me how to become a successful as well as noble person. I will not forget that it was his encouragements and supports that helped me through my hard time in 2006, when I was diagnosed with a tumor. He has set a paradigm for me in my future academic career. I would like to extend appreciation to my colleagues through all these years, in particular, Dr. Damien Rodger, for his long time collaboration and generous help to me. My gratitude also goes to all former and current students of the Caltech Micromachining Laboratory: Fukang Jiang, Ellis Meng, Jun Xie, Justin Boland, Qing He, Matthieu Liger, Victor Chi-Yuan Shih, Scott Miserendino, Changlin Pang, Angela Tooker, Siyang Zheng, Po-Jui Chen, Nick Lo, Brandon Quach, Jason Shih, Mike Liu, Luca Giacchino, Ray Huang, Jeffrey Lin, Monty Nandra, Justin Kim, Bo Lu, Yok Satsanarukkit, Wendian Shi, Ying-Ying Wang and Parvathy Menon. It is my true pleasure to have carried many

v valuable discussions with all of you and have you as my lifelong friends. I would also like to acknowledge Mr. Trevor Roper, Mesdames Tanya Owen, Christine Garske, and Agnes Tong for their cordial assistance in lab work and administrative jobs. The interdisciplinary nature of my work has also provided me the opportunity to work with many wonderful collaborators from the Keck School of Medicine of the University of Southern California and from the University of California, Santa Cruz. In particular, Prof. James Weiland, Prof. Mark Humayun and Dr. Hossein Ameri have taught me a great deal of medical knowledge and provided surgical assistance for my research. Also I would like to thank Prof. Gerald Loeb and Dr. Jesse Singh, Prof. Wentai Liu, Dr. Mohanasankar Sivaprakasam, and Yang Zhi for helping with some device design and experiments. My gratitude also goes to my Ph.D. candidacy and defense committee members, including Prof. Yu-Chong Tai, Changhuei Yang, James Weiland, Babak Hassibi, Azita Emami, and Joel Burdick, for their helpful and insightful input to my work. Finally, I would like to thank my parents, my elder brother, and my fiancé, Guangxi Wang, to whom this thesis is dedicated, for their unconditional love, support, and company through my best and worst time. They are the greatest source of strength in my life.

vi ABSTRACT

Integrated wireless implants have always been the ultimate goal for neural prostheses. However, technologies are still in development and few have actually been transferred to clinical practice due to constraints in material biocompatibility, device miniaturization and flexibility. In this dissertation, emphasis is placed on the development of Parylene devices for neural prostheses, and particularly, for retinal prostheses that partially restore lost vision for patients suffering from outer retina degeneration. A basic Parylene-metal-Parylene skin technology for making planar Parylene micro-electro-mechanical systems (MEMS) devices, such as electrode arrays and radiofrequency (RF) coil, is first discussed, followed by accelerated lifetime soaking tests to investigate the long term stability of such skins in hot saline under both passive and active electrical stressing. Discussion is further expanded on a detailed description of the design, fabrication, and testing procedure of two types of MEMS coils, which serve as receiver coils for wireless power and data transfer in a retinal implant system. After that, an embedded chip integration technology is presented, which allows the integration of complementary metal-oxide-semiconductor (CMOS) integrated circuit (IC) chips with other MEMS devices and discrete components so as to achieve high-level system functionality. Finally, an integrated wireless neural stimulator is designed and successfully fabricated using a test chip.

vii TABLE OF CONTENTS

  Acknowledgements ··············································································································· iv  Abstract ································································································································· vi  Table of contents ·················································································································· vii  List of illustrations ················································································································ xi  List of tables ························································································································ xix  Nomenclature ······················································································································· xx  C h a p t e r 1 Introduction ······································································································ 1  1.1 Outer retinal diseases ··································································································· 1  1.2 Artificial visual prostheses for AMD and RP treatments············································· 3  1.2.1 Introduction to visual prostheses ........................................................................ 3  1.2.2 State-of-the-art for epiretinal prostheses .......................................................... 10  1.2.3 Technical challenges of advanced epiretinal prostheses .................................. 12  1.3 Parylene-based MEMS technologies ········································································· 13  1.3.1 Introduction to MEMS ...................................................................................... 13  1.3.2 Introduction to Parylene C ................................................................................ 16  1.3.3 Parylene-based MEMS for neural prostheses .................................................. 19  1.4 Goal and layout of the dissertation············································································· 21  1.5 Summary ···················································································································· 22 

viii C h a p t e r 2 Parylene-metal-Parylene skin and its Long-term biostability ························ 23  2.1 Introduction ················································································································ 23  2.2 Parylene-metal-Parylene skin technology ·································································· 23  2.2.1 Review of flexible MEMS skin technologies .................................................. 23  2.2.2 Process development ......................................................................................... 25  2.2.3 Adhesion issue on interfaces............................................................................. 27  2.3 Interfacial adhesion enhanced process ······································································· 28  2.3.1 Accelerated-lifetime soaking test of Parylene bi-layer structures ................... 28  2.3.2 Water vapor permeation through thin-film Parylene C ................................... 32  2.3.3 Parylene heat molding....................................................................................... 35  2.4 Accelerated-lifetime soaking test of Parylene-metal-Parylen skins··························· 36  2.4.1 Sample preparation ........................................................................................... 36  2.4.2 Passive soaking test of Parylene-protected metal ............................................ 37  2.4.3 Active soaking test of Parylene-protected metal .............................................. 40  2.5 Summary ···················································································································· 45  C h a p t e r 3 Design of Parylene-based mems coils···························································· 46  3.1 Introduction ················································································································ 46  3.2 Inductive link system overview ················································································· 46  3.3 Design of planar MEMS coils···················································································· 49  3.3.1 Self-inductance .................................................................................................. 49  3.3.2 Effective series resistance ................................................................................. 50 

ix 3.3.3 Parasitic capacitance ......................................................................................... 52  3.3.4 Self-resonant frequency and Quality factor...................................................... 55  3.3.5 Simulation results .............................................................................................. 56  3.4 Inductive power link ·································································································· 58  3.5 Summary ···················································································································· 68  C h a p t e r 4 Implantable MEMS RF coil for power and data trasmission························· 69  4.1 Introduction ················································································································ 69  4.2 Planar MEMS RF coil ································································································ 69  4.2.1 Design and fabrication ...................................................................................... 69  4.2.2 Fabrication results ............................................................................................. 71  4.2.3 MEMS coil characteristics ................................................................................ 72  4.2.4 Data and power transfer measurements............................................................ 75  4.3 Fold-and-bond MEMS coil ························································································ 77  4.3.1 Introduction ....................................................................................................... 77  4.3.2 Fabrication ......................................................................................................... 78  4.3.3 Fabrication and testing results .......................................................................... 81  4.4 Summary ···················································································································· 87  C h a p t e r 5 Implantable RF-coiled chip integration technology ······································· 89  5.1 Introduction ················································································································ 89  5.2 Accelerated-lifetime soaking test of Parylene protected ICs ····································· 90  5.3 Integration demonstration ·························································································· 94 

x 5.3.1 Design and fabrication ...................................................................................... 94  5.3.2 Fabrication and functional test results .............................................................. 99  5.4 Mechanical model for implantation studies ····························································· 101  5.5 Test chip system integration····················································································· 108  5.5.1 Functional system prototype ........................................................................... 108  5.5.2 Fabrication of the assembled BION system ................................................... 110  5.5.3 Functional test of the assembled BION system ............................................. 115  5.5.4 Monolithic system processing ........................................................................ 118  5.6 Summary ·················································································································· 120  Appendix A: Dual-layer RF MEMS coil microfabrication process··································· 122  Appendix B: Fold-and-bond RF MEMS coil microfabrication process ···························· 128  Appendix C: Embedded chip integration microfabrication process ·································· 132  Bibliography······················································································································· 143 

xi LIST OF ILLUSTRATIONS

Figure 1.1: (a) Normal vision. (b) Loss of central vision in patients with AMD. (c) Tunnel vision in patients with RP. (Images courtesy of Artificial Retina Project) ............................. 2  Figure 1.2: Horizontal cross section through human eye and the cell layers in the retina. (Eye illustrate adapted from Bausch & Lomb) ........................................................................ 3  Figure 1.3: Two commercially available neural prosthesis products: (a) A single channel cochlear implant (Image adapted from Med-EL medical electronics). (b) A single channel pacemaker. (Image adapted from Wikimedia Commons)....................................................... 4  Figure 1.4: A blind patient with a cortical prosthesis system for over 20 years. The right picture shows the implanted 64-electrode array. The wires are connected to electrodes through a burr hole.................................................................................................................... 5  Figure 1.5: Several types of microelectrodes for cortical prostheses: (a) Utah 100-channel electrode array [17]. (b) Michigan 16-channel electrode array with flexible silicon ribbon cable [18]. (c) Caltech 2-D 32-channel electrode array with flexible Parylene cable [19]. ... 5  Figure 1.6: Optical nerve prosthesis systems: (a) Approach used spiral cuff electrodes [20]. (b) “Hybrid retinal implant” approach [21]. ............................................................................ 7  Figure 1.7: Subretinal prostheses. (a) Schematic showing the location of the implants. (b) The microphotodiode array made by Zrenner et al [24]. (c) The 5000 microphotodiode array made by Chow et al [25]. ................................................................................................ 8  Figure 1.8: System overview of the epiretinal prosthesis. (Illustration courtesy of NextGenLog) .......................................................................................................................... 10 

xii Figure 1.9: Several epiretinal implants: (a) Implanted Second Sight 16 electrode array [27]. (b) Retinal stimulator developed by Intelligent Medical Implants (Image adapted from Intelligent Medical Implants). (c) Epiretinal stimulator made by the German team. (d) Test platform of an epiretinal prosthesis system (Rizzo and Wyatt) (Image adapted from The Boston Retinal Implant Project). ............................................................................................ 11  Figure 1.10: Illustration of bulk micromachining (A) and Bosch process (B). .................... 15  Figure 1.11: Illustration of surface micromachining. ............................................................ 15  Figure 1.12: Chemical structures of Parylene N, C, D, and HT, and some of the correlated process parameters used in Parylene deposition system PDS 2010 and 2060...................... 17  Figure 1.13: Typical morphology of rabbit retinas after six month implantation. (Images courtesy of Dr. Damien Rodger) ............................................................................................ 19  Figure 1.14: Parylene-based devices for neural prostheses: (a) cochlear probe; (b) 3-D neural probe; (c) retinal multielectrode array; (d) spinal cord electrodes. ............................ 20  Figure 1.15: Proposed all-intraocular system for high density epiretinal implant. [35] ....... 21  Figure 2.1: Illustration of the flexible MEMS skin technology. ........................................... 24  Figure 2.2: Conceptual illustration of Parylene-metal-Parylene skins.................................. 25  Figure 2.3: Fabrication process of Parylene-metal-Parylene skin technology. .................... 26  Figure 2.4: (a) Untreated Parylene bi-layer sample after one day soaking. Water bubbles are formed at the interface. (b) Annealed Parylene-bi-layer sample after long-term soaking. No delamination or water bubble is observed........................................................................ 28  Figure 2.5: Testing cells of Parylene bi-layer samples are placed in a convection oven. .... 30 

xiii Figure 2.6: Examples of Parylene heat molding: (left) Different shapes are formed for Parylene films. (right) Demonstration of flexibility after heat molding. .............................. 35  Figure 2.7: A Parylene packaged thin film resistor with pure gold metallization. ............... 36  Figure 2.8: Microscope images of samples before and after passive ALSTs, showing different corrosion mechanisms. ............................................................................................ 38  Figure 2.9: Simplified electrochemical test setup for active soaking test. A tested device is shown in the right picture. ...................................................................................................... 41  Figure 2.10: Typical aging curve of a sample: (top) from sample set #3. (middle) from sample set #6. (bottom) from sample set #9. ......................................................................... 41  Figure 2.11: Failure procedure observed in sample with 4.7 μm Parylene coating. ............ 42  Figure 2.12: Typical SEM images and EDS spectra on the top side and the back side of the breakdown area. ...................................................................................................................... 43  Figure 2.13: Even bubbles observed on samples from set #9 after soaking in saline at 90 ºC for 65 days............................................................................................................................... 44  Figure 3.1: System configuration for the proposed system [64]. .......................................... 47  Figure 3.2: Average SAR for an average man exposed to 1mW/cm2 plane wave. (E polarization denotes the electric field parallel to the body, H denotes the magnetic field parallel to the body and K denotes the wave moving from head to toes or toes to head.) [70] ................................................................................................................................................. 49  Figure 3.3: Simplified representation of a circular planar MEMS coil................................. 49  Figure 3.4: Voltage profile of an n-turn planar coil. [74] ...................................................... 53  Figure 3.5: Voltage distribution of an N-layer coil. ............................................................... 54 

xiv Figure 3.6: Equivalent RLC circuit of a MEMS coil. ............................................................ 55  Figure 3.7: Simulated self-inductance and ESR of the sample coil. (Simulation courtesy of Dr. Wen-Cheng Kuo.) ............................................................................................................ 58  Figure 3.8: Simulated self-inductances and ESR as functions of trace widths. The Qfactors are summarized in the table. (Plots courtesy of Dr. Wen-Cheng Kuo.) .................. 58  Figure 3.9: System overview of an inductive power link for biomedical applications. The secondary stage is modeled as a nonlinear circuit. ................................................................ 59  Figure 3.10: The secondary stage is simplified with an approximated linear model. .......... 59  Figure 3.11: Depiction of the inductive link, showing the size and relative position of the transmitter coil and receiver coils. ......................................................................................... 63  Figure 3.12: Coupling coefficient as a function of coil separation (Transmitter coil ~ 30.8 mm in OD. Receiver coil ~ 9mm in OD).............................................................................. 64  Figure 3.13: Analytical simulations of voltage gain (top) and power gain (bottom) as functions of the receiver coil geometries. (n2 denotes the number of turns per layer.) ....... 66  Figure 3.14: Power delivered to the load, power dissipation and coil Q factors vs. number of layers. (a) Coil with an inductance of 5.47 μH. (b) Coil with an inductance of 9.73 μH. ................................................................................................................................................. 68  Figure 4.1: Schematics of a dual-metal-layer MEMS coil. ................................................... 69  Figure 4.2: Detailed process of Parylene-based MEMS coil fabrication. ............................. 70  Figure 4.3: A fabricated ocular coil sitting on a penny. The right microscope image shows the interconnection via between two layers of metals. .......................................................... 72  Figure 4.4: Demonstration of coil flexibility and foldability. ............................................... 72 

xv Figure 4.5: Equivalent RLC circuit of a MEMS coil. ............................................................ 73  Figure 4.6: Typical I-V curve of the Parylene-based coil. .................................................... 74  Figure 4.7: Coil impedance measurement and curve fitting using the 3-element model: (right) Imaginary part; (left) Real part. (Data correspond to electrical parameters of coil in Figure 4.3, Ls = 1.19 μH, Rs = 72 Ω and Cs = 201 pF). ..................................................... 75  Figure 4.8: Schematic of the experimental setup for data transmission measurement. (Images courtesy of Dr. Wentai Liu) ..................................................................................... 76  Figure 4.9: Inductive coupling test waveforms: (left) received signal is 25 mV peak to peak; (right) received signal is 15 mV peak to peak. (Plots courtesy of Dr. Wentai Liu) ............. 76  Figure 4.10: Concept of fold-and-bond technology for Q-factor enhancement. A coil with one fold is depicted for representation. .................................................................................. 78  Figure 4.11: (a) Fabricated dual-metal-layer Parylene-based skins; (b) Microscope image of an interconnection via between two metal layers; (c) Photos of device details (From left to right): conductive wires of the coil, folding junction and suturing holes. ............................ 79  Figure 4.12: Thermal bonding results of: (a) Sample annealed at 200 ºC shows delamination; (b) Good sample annealed at 250 ºC. ............................................................. 80  Figure 4.13: Flexibility demonstration of the folded-and-bond devices. .............................. 81  Figure 4.14: Stretching marks along the folding line. ........................................................... 81  Figure 4.15: (a) Fold-and-bond coils after thermal bonding. (b) Overlapping metal wires with misalignments of 10 μm to 30 μm. ................................................................................ 82  Figure 4.16: Q-factors as a function of frequencies............................................................... 84  Figure 4.17: Experimental setup for power transmission measurement. ............................. 85 

xvi Figure 4.18: Power transfer efficiency at 1 MHz vs. separation distance of the coil pair.... 86  Figure 4.19: Power transfer efficiency vs. separation distance of the coil pair. ................... 87  Figure 5.1: Microscope images of chip metal pads. (a) A chip coated with annealed Parylene C. (b) A chip with regular Parylene C after 6 months of soaking. (c) A bare chip after 2 days of soaking. ........................................................................................................... 91  Figure 5.2: (left) Assembled RFID chip soaked in saline. (right) Close-up image of the attached RFID chip. ................................................................................................................ 92  Figure 5.3: Test setup for the active soaking test of RFID chips. ......................................... 92  Figure 5.4: Active soaking test results of Parylene protected IC chips................................. 93  Figure 5.5: Parylene cracks is sample with (left) 11.7 μm Parylene coating and (right) 22.1 μm Parylene coating. .............................................................................................................. 94  Figure 5.6: Concept of embedded chip integration. .............................................................. 95  Figure 5.7: (left) EM 4100 RFID chip used to demonstrate the integration technology. (right) WYKO image of a typical chip. ................................................................................. 95  Figure 5.8: Detailed process flow for embedded chip integration. ....................................... 97  Figure 5.9: Metal pole with a silicon piece for use to push the chip to the silicon shell. ..... 97  Figure 5.10: Vertical displacements of chips: 0μm step height indicates wafer level. ........ 98  Figure 5.11: (a) Example of 10 μm lateral misalignment of the chip. ...................................................................................... 99  Figure 5.12: A fabricated device: (a) Overall view of the device. (b) Close-up view of the embedded chip. (c) Close-up view of the coil wires. .......................................................... 100  Figure 5.13: Flexibility of the Parylene-metal-Parylene thin film. .................................... 100 

xvii Figure 5.14: Test setup to verify the function of embedded RFID chips............................ 101  Figure 5.15: Signal readout from the oscilloscope. ............................................................. 101  Figure 5.16: Prototype geometry for an all-intraocular Parylene-based device with all required component regions for a completely implantable system. .................................... 103  Figure 5.17: Custom heat-forming mold. (Images courtesy of Dr. Damien Rodger) ........ 103  Figure 5.18: Two surgical models for acute implantation: model-I (a) is designed for test chip integration; model-II (b) is designed for IRP1K integration. ...................................... 104  Figure 5.19: Example surgical photographs of prototype geometry implantation using (a) test chip surgical model; and (b) IRP1K surgical model. .................................................... 105  Figure 5.20: Improved geometry for a completely implantable system. ............................ 106  Figure 5.21: Example of a coil with ACIOL haptics attached. ........................................... 106  Figure 5.22: Two surgical models for chronic implantation in canine eyes. ...................... 107  Figure 5.23: Surgical implantation photographs of improved surgical model under: (left) anterior illumination, and (right) intravitreal illumination. ................................................. 107  Figure 5.24: BION chip layout and the pad connection. ..................................................... 109  Figure 5.25: Schematic overview of a BION chip integration system. ............................. 110  Figure 5.26: Fabrication process for the Parylene-based carrier substrate. Steps (b)-(d) describe the lift-off technology for platinum patterning...................................................... 111  Figure 5.27: (left) Fabricated carrier substrate. (right) Parylene ribbons to hold the chip. ............................................................................................................................................... 112  Figure 5.28: A fabricated fold-and-bond coil with two layers of metal. The electrical characteristics are measured and given in the table. ............................................................ 112 

xviii Figure 5.29: (a) A assembled BION system. (b) Interconnects formed with biocompatible conductive silver epoxy: (left) coil contacts, (right) chip and capacitor contacts............... 113  Figure 5.30: Examples of chip interconnections. For the right sample, short circuit is formed between adjacent pads. ............................................................................................ 114  Figure 5.31: A telemetry setup for functionality test of the assembled BION system. The right picture shows a personal trainer and other peripheral accessories. ............................ 116  Figure 5.32: Typical simulation pulse measured from the electrode sites. ........................ 117  Figure 5.33: Waveforms of transferred voltage and current to the BION circuitry. ......... 117  Figure 5.34: Measured stimulating pulses (a), transferred voltages (b) and transferred power (c) at different separation distances between two coils. ........................................... 118  Figure 5.35: Integration process for monolithic neural stimulator...................................... 119  Figure 5.36: Photoresist cracking and Parylene delamination encountered during the monolithic integration processing. ....................................................................................... 120 

xix LIST OF TABLES

Table 1.1: Properties of Parylene N, C, D, HT, and PDMS. (Table adopted from [36]) ..... 17  Table 2.1: Samples preparation of Parylene bi-layers for soaking tests. .............................. 29  Table 2.2: Accelerated-lifetime soaking test results of Parylene bi-layer samples. Red color denotes failed samples. Blue color denotes good samples. .................................................. 30  Table 2.3: Published WVTR data for Parylene C. (Courtesy of Parvathy Menon).............. 33  Table 2.4: Average WVTR of Parylene C thin films prepared under different conditions. Each sample set contains 2 samples from a same batch of Parylene deposition. (Data courtesy of Menon Parvathy[57].) ......................................................................................... 34  Table 2.5: Sample preparations. ............................................................................................. 37  Table 2.6: Soaking test results for Parylene-metal-Parylene structures under passive conditions, and extrapolated data according to the Arrehnius relationship. ......................... 39  Table 3.1: Comparison of available power sources [64]. ...................................................... 47  Table 3.2: Design specifications of a sample coil. ................................................................ 57  Table 4.1: Design specifications of the fold-and-bond coils. ................................................ 82  Table 4.2: Extracted electrical parameters of fold-and-bond coils. ...................................... 83 

xx NOMENCLATURE

MEMS: Micro-electro-mechanical systems RF: Radio-frequency CMOS: Complementary metal-oxide-semiconductor IC: Integrated circuit AMD: Age-related macular degeneration RP: Retinitis pigmentosa WHO: World Health Organization USC: University of Southern California MST: Microsystems technology DRIE: Deep reactive-ion etching LIGA: German acronym for “X-ray lithography, electroforming, and molding” HNA: Hydrofluoric acid & Nitric acid & Acetic acid TAMH: Tetra-methyl-ammonium-hydroxide PDMS: Poly(dimethylsiloxane) USP: United States Pharmacopeia ASIC: Application-specific integrated circuit ALST: Accelerated-lifetime soaking test

xxi RIE: Reactive ion etch MTTF: Mean-time-to-failure WVTR: Water vapor transmission rate WVP: Water vapor permeability SEM: Scanning electron microscopy EDS: Energy dispersive X-ray spectroscopy SAR: Specific absorption rate ESR: Effective series resistance UCSC: University of California, at Santa Cruz MFI: MicroFlex interconnection CL-I2: Chip-level integrated interconnect RFID: Radio-frequency identity TTL: Transistor-transistor logic ACIOL: Chamber intraocular lens HMDS: Hexamethyldisilazane BHF: Buffered hydrofluoric acid IPA: Isopropanol

1 C h a p t e r 1 INTRODUCTION

1.1 Outer retinal diseases Age related macular degeneration (AMD) and retinitis pigmentosa (RP) are two of the most common outer retinal degenerative diseases that result in human vision impairment and blindness. According to the World Health Organization (WHO), AMD has become the third leading cause of blindness on global scale, and ranks first in developed countries [1]. In fact, it has been estimated that almost 3 million people in the United States will have significant symptoms associated with AMD by 2020 [2]. As for RP, it has affected more than 500,000 people in the United States, among whom 20,000 are legally blind [3]. AMD primarily affects the central vision regions in people age 60 and older (Figure 1.1 (b)). Because the brain will compensate for central dark patches during its early stages, it is difficult to diagnose this disease until the advanced stages. Whereas for patients with RP, vision deterioration develops gradually and progressively, starting from defective dark adaptation (night blindness), followed by the reduction of peripheral vision (known as tunnel vision in Figure 1.1 (c)), and, sometimes, a complete vision loss in the advanced stages of the disease [4].

2

Figure 1.1: (a) Normal vision. (b) Loss of central vision in patients with AMD. (c) Tunnel vision in patients with RP. (Images courtesy of Artificial Retina Project) Although the mechanisms of these retinal diseases are not fully understood, histopathological studies indicate they are mainly restricted to outer retinal region, and associated with the damage of photoreceptor cells [5-8]. The photoreceptor cells (rods and cones) are situated in a deep retinal layer, and they, along with other cells (bipolar, ganglion, horizontal and amacrine cells), perform significant visual processing and communication tasks in normal vision (Figure 1.2). In a healthy retina, particularly, the photoreceptors initiate neural responses to incoming light by converting light into chemical and electrical signals. The bipolar and ganglion cells in the inner retina then pass the visual signals toward the optic nerve. Along the way, the horizontal and amacrine cells provide lateral interaction between neighboring cells [9]. Current treatments for such retinal degenerative diseases rely on gene replacement therapy, pharmaceutical therapy, nutritional therapy and dietary to slow down the development of blindness in their early stages [10, 11]. Nevertheless, once the vision is completely lost, no effective treatment exists that can cure these diseases. Studies on animal models show that photoreceptor and stem cell transplantations could be possible treatments by replacing the damaged retinal cells [12, 13]. However, they are still in

3 development, and not yet ready for clinical practice. Other concerns such as expensive costs and political issues also add barriers to their implementation. An alternative approach is proposed as bioelectronic visual prostheses, which bypass the damaged photoreceptor cells and utilize electrical stimulating to restore vision.

Figure 1.2: Horizontal cross section through human eye and the cell layers in the retina. (Eye illustrate adapted from Bausch & Lomb) 1.2 Artificial visual prostheses for AMD and RP treatments 1.2.1 Introduction to visual prostheses Efforts to develop neural prostheses are not new. In fact, intensive studies have been done since the 18th century, using electronic devices to stimulate neurons and muscles for the control of limbs and organs, or to elicit sensory sensation for feeling, hearing or seeing. For instance, cochlea implants and pacemakers, the best-known neural prostheses, have been accepted in clinical practice in recent decades (Figure 1.3).

For visual

prostheses, investigations on implantation devices began from the 1960’s. Since then,

4 many different approaches have been developed to restore lost vision function for blind patients, and each of them has distinct advantages and disadvantages.

Figure 1.3: Two commercially available neural prosthesis products: (a) A single channel cochlear implant (Image adapted from Med-EL medical electronics). (b) A single channel pacemaker. (Image adapted from Wikimedia Commons) I.

Visual cortex prosthesis Early work toward visual prostheses started from electrical stimulation of visual

cortex. In 1968, Brindley and Lewin implanted 80 platinum electrodes on the visual cortex of a 52-year-old blind patient though a burr hole [14]. They found that when excited by appropriate radio signals, the patient could experience sensations of light (phosphenes). Inspired by this work, Dobelle and Mladejovsky continued their investigation on 37 human subjects [15]. In 2000, Dobelle reported a subject who had been wearing a cortical visual prosthesis system for over 20 years [16]. This result demonstrated the clinical feasibility and stability of such cortical prosthesis system for the long-term implantation (Figure 1.4). However, the early studies used large electrodes and thus had drawbacks including flickering phosphene, interactions between phosphenes, limited resolution, pain, local

5 heating and electrolysis [9]. To overcome these problems, modifications to the implant systems, such as the development of microelectrode arrays (Figure 1.5), have been made in recent years. These modifications enabled devices to be embedded in the visual cortex (intracortical implants) and therefore higher resolution and lower stimulation current could be achieved.

Figure 1.4: A blind patient with a cortical prosthesis system for over 20 years. The right picture shows the implanted 64-electrode array. The wires are connected to electrodes through a burr hole.

Figure 1.5: Several types of microelectrodes for cortical prostheses: (a) Utah 100-channel electrode array [17]. (b) Michigan 16-channel electrode array with flexible silicon ribbon cable [18]. (c) Caltech 2-D 32-channel electrode array with flexible Parylene cable [19]. The main advantage of the cortical stimulation approach is that the skull can provide good protections for implanted devices.

In addition, the technique bypasses

diseases proximal to the primary visual cortex, making it potentially useful for treating the largest number of blind patients suffering from both optic nerve and retina damages.

6 However, this technique has inevitable disadvantages. First of all, spatial organization is more complicated at the cortical level and two adjacent stimulation sites are not necessarily located at adjacent areas in space, so that patterned electrical stimulations may not produce the same perceptional patterns.

Additionally, the surgical procedure is difficult and

complicated, which can have devastating results, including death. Finally, local heating is still a problem due to poor heat dissipation inside the brain. II.

Optic nerve prosthesis Optic nerve prosthesis uses direct optic nerve stimulation to restore vision. In one

method, spiral cuff electrodes were placed around the optic nerve of blind patients (Figure 1.6 (a)). Electrical stimulus applied to the optic nerve produced localized, sometimes colored phosphenes throughout the visual field. By varying the duration and amplitude of the stimulus, patients were able to perceive different brightness of phosphene [20]. Another method called “hybrid retinal implant” was conducted in recent years [21], which involves culturing neurons on micro-electro-mechanical systems (MEMS).

MEMS

components were connected with the central nervous systems (CNS) though cultured neurons and their extended axons. By this means, electrical stimulation from the electrodes will excite the cultured neurons to send action potentials toward the CNS neurons, and thus restore vision. Optic nerve prostheses can potentially be useful for both retinal degenerative and non-retinal degenerative blindness, without requiring surgical penetration of sensitive intracranial tissues. However, there are major problems associated with this approach: difficulties with high density stimulation and detail perception due to the high density of

7 the axons; surgical complexity and possible harmful side effects after implantation; and finally the difficulty to implement a complete intraocular system.

Figure 1.6: Optical nerve prosthesis systems: (a) Approach used spiral cuff electrodes [20]. (b) “Hybrid retinal implant” approach [21]. III.

Retinal prosthesis The retinal prosthesis approach aims to use the remaining visual pathway

component (optic nerve bipolar and ganglion cells) to provide partial restoration of vision. In the 1990’s, Humayun et al. conducted morphometric analysis of the retina of RP patients. They discovered that although the retinal photoreceptors are defective in those patients, their optic nerve, bipolar and ganglion cells to which the photoreceptors synapse remain functioning at a large rate [22]. Further studies showed similar results in AMD patients [23]. These results created profound impact on the ophthalmology field, by providing the possibility of using retinal prostheses to treat AMD and RP patients. Two types of prostheses are currently in development according to the layer of retina receiving the implanted device: subretinal and epiretinal. The subretinal prosthesis is performed behind the retina between the bipolar cell layer and the retinal pigment epithelium, by replacing the degenerated photoreceptors with

8 electronic devices (Figure 1.7). Microfabricated photodiodes are implanted either in the subretinal space through an incision in the sclera or by standard vitrectomy followed by retinotomy. The microphotodiodes are stimulated by incident light through the retina, and the resulting electric current excites adjacent sensory neurons. Because the devices are placed closer to the bipolar cells, small stimulation currents (50 μA) were reported to be sufficient to excite the neuron cells [24]. Clinical trials have been done by implanting arrays with 5000 microphotodiodes into right eyes of 6 patients with RP [25]. Visual function improvement occurred in all subjects, with no significant safety-related adverse effects (implant rejection, infection, retinal detachment, etc.) being observed after up to 18 months. This result demonstrates the safeness and efficacy of such subretinal prosthesis approach in short- term implantation.

Figure 1.7: Subretinal prostheses. (a) Schematic showing the location of the implants. (b) The microphotodiode array made by Zrenner et al [24]. (c) The 5000 microphotodiode array made by Chow et al [25]. Although the subretinal prosthesis shows advantages including lower stimulus threshold, high–density stimulation, and perhaps better heat dissipation through choroid,

9 there are still many problems with the current implant system. The main disadvantage is the disruption to the nutritional supply of the retina derived from the choroid due to the introduced barrier between the outer retina and the choroid. In addition, only a small amount of light can reach the microphotodiodes through retina, and therefore the efficiency of the system is dramatically reduced. The long-term stability of the subretinal prosthesis has not yet been verified, which is also a big concern for its final implementation. An epiretinal prosthesis system usually employs a multielectrode array implanted on the surface of the inner retina between the vitreous and internal limiting membrane. A data acquisition system located outside of the body captures images from the surroundings, and converts the information into patterns of electrical signals. Upon the reception of signals through data transmission and processing systems, the electrodes stimulate the remaining retinal ganglion cells and restore vision (Figure 1.8). A major concern with epiretinal implantation is that the implants will be exposed to ocular rotational movements, raising the need for a reliable attachment method to fix the electrode array on the retina. Despite this drawback, we believe epiretinal implantation is the most promising technology for treating AMD and RP diseases among all visual prosthesis approaches, because of its many advantages compared to others. First, the epiretinal implantation and follow-up examination only require standard ophthalmologic technologies, reducing the risk of trauma during the surgery as well as allowing for the implant to be replaced. In addition, most of the implanted electronics is kept in the vitreous cavity, such that the implant can take the advantage of the heat dissipating properties of the vitreous, and thus causes less damage to the retina compared with subretinal implants.

10 Developments of MEMS technologies provide the possibility of complete intraocular retinal implants and high-density array stimulation.

Figure 1.8: System overview of the epiretinal prosthesis. (Illustration courtesy of NextGenLog) 1.2.2 State-of-the-art for epiretinal prostheses In 1994, Humayun et al. reported the results of electrically stimulating vertebrate retinas, showing that surface electrical stimulation of the inner retina in eyes with outer retinal degeneration can elicit a localized retinal response [26]. Since then, his team at the Doheny Eye Institute at the University of Southern California (USC), collaborating with a group of researchers, has developed an extraocular retinal prosthesis technology. The implanted unit of the first generation model consists of an intraocular part (electrode array) and an extraocular part (electronic box including a receiving coil).

A 16 platinum

microelectrode array, made by Second Sight Medical Products, Inc (Sylmar, CA, USA), is positioned temporal to the fovea and is attached to the inner retinal surface using a single tack. A hand-made coil is implanted outside of the eye, in the temporal skull to transfer

11 power and data through a telemetry link. After demonstrating the device functionality in animal models and short-term human studies, clinical trials of the IRP was conducted in 2002 to verify the chronic, long-term feasibility of the implants. To date, 6 patients with bare light perception have received the implants. Current results show that subjects can distinguish motion and locate objects (plate, cup, and knife). They have also shown the ability to discriminate between percepts created by different electrodes based on their locations [27]. Another clinical trial of an epiretinal implant, featuring a stimulation array of 49 platinum electrodes on a polyimide substrate, has recently been performed by Intelligent Medical Implants AG (Zurich, Switzerland). Subjects with implants can see phosphenes and crude shapes correlating to applied stimuli [28]. Several other groups worldwide have developed different designs of epiretinal implants that vary in terms of system configurations. All these implants are guided by similar requirements, including preserving as much of the normal visual function as possible while minimizing the amount of implanted electronics. Their functionality and short-term stability have been confirmed in both animal models and human subjects. However, clinical trials are still needed to demonstrate the long-term performances of implants.

Figure 1.9: Several epiretinal implants: (a) Implanted Second Sight 16 electrode array [27]. (b) Retinal stimulator developed by Intelligent Medical Implants (Image adapted from Intelligent Medical Implants). (c) Epiretinal stimulator made by the German team.

12 (d) Test platform of an epiretinal prosthesis system (Rizzo and Wyatt) (Image adapted from The Boston Retinal Implant Project). 1.2.3 Technical challenges of advanced epiretinal prostheses The next generation epiretinal prosthesis will require entirely intraocular implants to minimize surgical complexity as well as to reduce physical damages to patients during implantation. In this case, a receiving coil of the telemetry link, along with electronic circuitry, will be placed inside the eye in the anterior chamber or lens capsule. A flexible cable extends through the posterior chamber to reach the multielectrodes array attached on the inner retina surface.

Due to the limited space of the human eye (< 1 cm3),

miniaturization is demanded for such an implant system. MEMS technology, as described in the next section, is a great candidate. Most of the designs for intraocular retinal prostheses so far have been focusing on discrete devices, then using hand assembly for the system integration. However, a larger number of electrodes will be needed for advanced retinal prostheses to provide more information at higher spatial resolution in order to improve the quality of the image. A conservative estimate of pixel density will be 32×32 to enable blind individuals to better carry out activities of daily living [28]. Current interconnect and packaging technologies are labor-intensive, inefficient, and costly, limiting the achievable device resolution. A desired approach would be to build microfabricated devices in a process compatible way, while still sufficiently versatile to integrate other discrete components (application specific integrated circuits (ASICs), capacitors, etc.).

13 Furthermore, retinal implants will need to address the long-term biocompatibility issue of the electronics and the multielectrode array in the saline environment of human eyes. Thus a hermetic, biocompatible, flexible packaging technology will be necessary to protect such implanted components from the corrosive substances. To overcome this challenge, Parylene C is selected as the main structural and packaging material because of its many favorable properties which will be discussed in the next section. 1.3 Parylene-based MEMS technologies 1.3.1 Introduction to MEMS MEMS, also called micromachines (in Japan) or microsystems technology (MST) (in Europe), is a microfabricated system comprising both electrical and mechanical components. Typical dimensions of such systems range from 100 nm to millimeters. The beginning of MEMS field concurred with the invention of the transistor by John Bardeen and Walter Brattain in 1947. In 1959, Richard Feynman mapped out the tremendous potentialities for miniaturization of machines to the scientist community, in his famous talked titled “There’s plenty of room at the bottom.” Since then, much attention has been drawn to this field, and a wide variety of MEMS devices have been built, such as micromotor [29] and accelerometer [30]. In addition to the device and application development, processing technologies and materials used in MEMS are continually evolving. From the processing aspects, many basic MEMS techniques are borrowed or adapted from semiconductor fabrication, such as photolithography, oxidation, physical vapor deposition, chemical vapor deposition, dry plasma etching. Other techniques specifically for MEMS have also been developed, such

14 as deep reactive-ion etching (DRIE), LIGA (German acronym for “X-ray lithography, electroforming, and molding”), electroplating, and soft lithography. From the material aspects, alternative substrate, such as glass and polymers are introduced. The driving factor for this change is the demand for making devices that are more biocompatible and cheaper. In fact, after polymer materials were introduced a few decades ago, MEMS has become a key technology for interfacing integrated electronics and living cells for biomedical applications [31]. In order to build complicated devices for various applications, multiple basic process steps are combined in two general ways: bulk micromachining and surface micromachining [32]. Bulk micromachining, as illustrated in Figure 1.10-A, uses chemical etching or plasma etching to selectively remove substrate material with oxide or photoresist as masking films. Typical etching process can be either isotropic etching using HNA (hydrofluoric acid + nitric acid + acetic acid) and XeF2, or anisotropic etching using KOH, TMAH (tetra-methyl-ammonium-hydroxide) and plasma. Among these techniques, DRIE, particularly the Bosch process, is developed for creating very high-aspect ratio structures, and even vertical through trenches in substrates (Figure 1.10-B).

15

Figure 1.10: Illustration of bulk micromachining (A) and Bosch process (B). Surface micromachining, on the other hand, comprises a number of techniques to build microstructures from thin films previously deposited on substrates (Figure 1.11). General process involves a sacrificial layer technique.

Common sacrificial materials

include silicon oxide, metal, and photoresist, which can be removed using chemicals. After a sacrificial layer is deposited and patterned on a substrate, structural thin film can be deposited and patterned, which will perform the mechanical or electrical function in the final device. Structural materials normally include silicon oxide, nitride, and polymer.

Figure 1.11: Illustration of surface micromachining.

16 Most devices described in this dissertation will use surface micromachining with Parylene C as the main structural material. Silicon only serves as a carrier, which will be completely removed after fabrication. System integration involves both surface and bulk micromachining technologies as the Bosch process would be used to etch through cavities for holding discrete components (capacitors, ASICs, etc.). 1.3.2 Introduction to Parylene C Parylene is the trade name for a family of polyxylylene polymers, which were discovered in 1947 and commercialized by Union Carbide Corporation in 1965 [33]. To date, four types of Parylene dimer are commercially viable: Parylene N, Parylene C, Parylene D, and Parylene HT. The deposition takes place at room temperature using a chemical vapor deposition process, known as Gorham process. Raw material dimer is placed in a vaporizer furnace, and evaporated to a dimeric gas under vacuum. The gas then passes through a high temperature pyrolysis furnace, where the molecule is cleaved into monomers. The monomers enter a room temperature deposition chamber, and polymerize on all exposed surfaces inside the chamber conformally [34]. The un-reacted monomers are collected to a cold trap before they enter the mechanical pump. Parylene coating thickness normally ranges from one to tens of microns, controlled by the amount of the dimer. Deposition rate is proportional to the monomer concentration, correlated with the deposition chamber pressure. Different deposition parameters are required for various types of Parylene, but the process steps essentially remain the same (Figure 1.12).

17

Figure 1.12: Chemical structures of Parylene N, C, D, and HT, and some of the correlated process parameters used in Parylene deposition system PDS 2010 and 2060. Table 1.1: Properties of Parylene N, C, D, HT, and PDMS. (Table adopted from [35])

The properties of different types of Parylene are summarized in Table 1.1, compared with poly(dimethylsiloxane)(PDMS).

It can be seen that Parylene, and

18 specifically, Parylene C, has many desirable properties for neural prostheses compared with two other commonly used materials, PDMS and polyimide [36]. It has pinhole-free comformal deposition, low gas and water vapor permeability, and chemical inertness, so that it can protect implanted devices and electronics from interacting with body fluids or interfering agents. Its good mechanical strength (elongation break of 200%) and low Young’s modulus (~4 GPa) make devices strong enough to survive surgical manipulations, and flexible enough to minimize damages to surrounding tissues. It is also optically transparent, allowing the anatomy to be seen though devices during implantation and follow-up. Most importantly, Parylene C is an ISO 10993 and United States Pharmacopeia (USP) Class VI biocompatible material, the highest biocompatibility and implant clearance among plastics. In fact, animal studies have been done to verify its biocompatibility in the eye environment. The experiments were conducted at Doheny Eye Institute of USC, by implanting untreated Parylene C thin films in the vitreous cavity of the right eyes of two rabbits. The thin films were left in the eyes for six months before the animal models were sacrificed. Post-mortem histological evaluation reveals no discernable difference between the retinas of experimental eyes and control eyes, indicating no detectable adverse immune response caused by implanted Parylene C (Figure 1.13). This result further strengthens the fact that Parylene C is biocompatible for intraocular retinal prostheses as well as other ocular implants.

19

Figure 1.13: Typical morphology of rabbit retinas after six month implantation. (Images courtesy of Dr. Damien Rodger) 1.3.3 Parylene-based MEMS for neural prostheses Parylene-C is originally used as an insulating and packaging material in the electronics industry. In recent years, there has been great interest in using Parylene-based MEMS for neural implant applications, where Parylene serves not only as a protecting layer, but also as a primary structural material. The main advantage of using Parylene in MEMS fabrication lies in its ease of integration with other microfabrication techniques.

The unique room-temperature

conformal deposition characteristics make it post-COMS compatible, so that electrodes, sensors, actuators, and even circuit can be incorporated into Parylene devices. Furthermore, Parylene can be easily patterned with oxygen plasma etching, using either photoresist or metal as masks. However, one major concern is the adhesion between Parylene and other materials commonly used in MEMS fabrication. It has been found that the adhesion of photoresist and Parylene is generally good; whereas some materials, such as silicon oxide, evaporated gold and aluminum do not adhere to Parylene very well. To address the adhesion problem, several techniques and adhesion promoters have been developed.

For instance, a saline-based adhesion promoter (A-174) can create good

20 bonding strength between Parylene and many substrates. Physical surface roughening using oxygen plasma, bromine trifluoride or Xenon difluoride can also help improve the dry adhesion between different layers of Parylene, as well as Parylene and silicon surfaces [37]. Consequently, devices with multi-layer Parylene structures can be implemented without special planarization techniques. A large variety of devices have been built for neural prostheses using Parylene micromachining, such as electrodes for cochlear implantation [38], Parylene cabled neural probes [39] for cortical implantation, flexible multielectrode arrays for retinal and spinal cord implantation [35]. More Parylene-based devices and technologies will be discussed in this dissertation.

Figure 1.14: Parylene-based devices for neural prostheses: (a) cochlear probe; (b) 3-D neural probe; (c) retinal multielectrode array; (d) spinal cord electrodes.

21 1.4 Goal and layout of the dissertation Our solution to the advanced retinal prostheses is to develop an all-intraocular system using Parylene-based MEMS technologies. The system, as shown in Figure 1.15, would comprise two MEMS radio-frequency (RF) coils for receiving wireless power and data signal, circuitry integrated on a flexible Parylene cable for converting the signal to simulation pulses, and a high-density MEMS electrode array for simulating the neural cells. The electrodes array has already been invented by Dr. Damien Rodger in the Caltech Micromachining laboratory [35]. Therefore, this work will center on MEMS Parylenebased technologies for RF coil fabrication and full system integration, which aim to contribute to the overall initiative of realizing neural prosthetic systems and particularly, an all-intraocular epiretinal implant system.

Figure 1.15: Proposed all-intraocular system for high density epiretinal implant. [35] Chapter 2 describes the fabrication technology and the long-term reliability of a Parylene-metal-Parylene skin, which is a basic structure for planar Parylene MEMS device, such as electrode arrays and coils. Chapter 3 discusses the design procedure of the MEMS

22 coil from a theoretical point of view using analytical models, followed by Chapter 4, where coil prototypes are experimentally designed, fabricated and measured.

Chapter 5

introduces a wafer level integration technology, enabling the integration of applicationspecific integrated circuits (ASICs) with other system components. We also present the detailed design, fabrication and functional testing of a full implantable, flexible, Paryleneenabled neurostimulator that features single channel wireless stimulation capability in Chapter 5. 1.5 Summary Two blindness diseases due to outer retinal degeneration are described in this chapter. Research shows that the visual prosthesis technology has great potential for treating these diseases and restoring the lost vision function. Four visual implantation approaches are discussed, among which the epiretinal retinal prosthesis has distinct advantages over others. However, a number of obstacles still exist, mainly related to material choice and fabrication development.

Therefore, an all-intraocular epiretinal

implant system is proposed to address these issues, utilizing Parylene-based MEMS technologies for device fabrication and system integration.

23 C h a p t e r 2 PARYLENE-METAL-PARYLENE SKIN AND ITS LONG-TERM BIOSTABILITY

2.1 Introduction As discussed in Chapter 1, one of the main goals of this work is to develop MEMS RF coils as receiver coils for wireless power and data transfer. A key technology for making such coils is so called a flexible Parylene-metal skin technology, featuring a Parylene-metal-Parylene sandwich structure. In this chapter, we will present the detailed process for making Parylene-metal-Parylene skins. The feasibility of using these skins for long-term implantation is also studied using an accelerated lifetime soaking test (ALST). 2.2 Parylene-metal-Parylene skin technology 2.2.1 Review of flexible MEMS skin technologies The flexible MEMS skin technology refers to a microfabrication technology that enables the integration of MEMS devices and ICs onto a flexible skin normally made of polymer materials. The concept was first proposed in 1985 by Barth et al., in which a onedimensional flexible Si-diode temperature sensor array was built using a polyimide strip to connect Si islands formed by isotropic HNA etching [40]. However, their technology has serious reliability problems and needs major improvements. In 1997, Jiang et al. reported the results of a remarkable work, where a reliable, large scale 2-D shear stress sensor skin was developed and successfully demonstrated for use in aerodynamic applications [41]. As a follow up of this study, an IC-integrated shear-stress sensor skin was invented, showing the capability of integrating CMOS circuitry on the flexible skin [42].

24 The fabrication process, as illustrated in Figure 2.1, uses polyimide as structural layers, connecting Si islands formed by selective wet etching (KOH or TMAH) or plasma dry etching (DRIE) [43]. Planar MEMS structures, such as electrode contacts and metal leads, will be directly fabricated on the polymer substrate. Si MEMS devices, such as ICs, Si sensors and actuators, will be processed on the Si islands.

Figure 2.1: Illustration of the flexible MEMS skin technology. A big advantage of this process is that it is compatible with other MEMS technologies so that only minimal adjustments are needed in order to build flexible MEMS skins.

However, this process is not applicable to retinal implants because the

biocompatibility of silicon is arguable although some evidences have shown that bare silicon can support cell growth [44]. Recent efforts on flexible MEMS skins use polyimide as substrate and completely remove the silicon substrate [45, 46], but they still need to tackle the long-term biocompatible difficulty of the polyimide. Therefore, it is necessary to develop a fully implantable MEMS skin technology and Parylene emerges as a good candidate because of its many desirable properties.

25 2.2.2 Process development Parylene-metal-Parylene skins will typically involve thin-film metal structures sandwiched by Parylene C, as shown in Figure 2.2. For simplicity, only one layer of metal is drawn to illustrate the concept. For actual devices in general, multiple layers can be achieved, which will be further discussed in Chapter 4.

Figure 2.2: Conceptual illustration of Parylene-metal-Parylene skins. A typical fabrication process for making such skins with one layer of metal is depicted in Figure 2.3. First, a layer of AZ4400 photoresist (AZ Electronic Materials, Branchburg, NJ, USA) is optionally spun on a standard silicon wafer as the sacrificial layer, and hard baked at 100ºC for overnight. After that, a layer of Parylene C is vapordeposited in a PDS 2010 system (Specialty Coating Systems, Indianapolis, IN, USA) on top of the photoresist, followed by metal deposition using an e-beam evaporator (SE600 RAP, CHA Industries, Fremont, CA, USA). An AZ 1518 layer is then spun on top of the metal, exposed in a 10× reduction GCA Mann 4800 DSW wafer stepper (General Signal Corporation, Stamford, CT, USA), or a Kasper 2001 contact aligner (Kasper Instruments, Inc., Sunnyvale, CA, USA), depending on the resolution requirement of metal features, and developed to form a photoresist mask for metal etching. After metal patterning, the

26 photoresist mask is removed, and another layer of Parylene C is deposited to seal the entire metal structure. A photoresist etch mask is then spun and patterned to expose the areas of the contact pads, as well as to define the device profile. Finally, unwanted Parylene is etched using oxygen plasma in a reactive ion etch (RIE) system (Semi Group Inc. T1000TP/CC), and the entire flexible structure is released from the silicon substrate by dissolving the photoresist sacrificial layer using a standard acetone etch process. If no sacrificial photoresist is applied, devices can be directly peeled from the natively oxidized silicon substrate in a water bath due to the hydrophobicity of the underlying Parylene C surface.

Figure 2.3: Fabrication process of Parylene-metal-Parylene skin technology. The Parylene-metal-Parylene skin technology serves an important role in the Caltech Micromachining Laboratory, and many devices have been developed using this technology, such as multielectrode array skins and flexible Parylene cables. For the retinal prosthesis project, this technology has its distinct advantages. First, by using Parylene directly as the actual substrate, both the fabrication and packaging procedures of the device can be greatly simplified. Additionally, the skins are flexible and foldable so that they can

27 be implanted through small surgical incisions, allowing wounds to heal quickly. Moreover, the metal lines are completely padded by the flexible Parylene material, and can therefore with stand repeated bending during surgical handling. Finally, a heat-molding treatment, which will be discussed in Section 2.3.3 , has been developed to modify the skins into various shapes, allowing them to easily match the curvature of the target implant area. 2.2.3 Adhesion issue on interfaces With the Parylene technology at hand, the next step is obviously to investigate its long-term reliability because human body, fairly well approximated by an oxygenated saline solution with a salt content of about 0.9% at pH ~ 7.4 and temperature of 37±1 ºC [47], is a harsh environment for metals. Although researches have been done to evaluate the biostability of Parylene packaging in the past years [48, 49], the long-term reliability of such Parylene-metal-Parylene sandwich structures has never been studied. Initial experiments to test the Parylene sealing behavior are carried out by soaking Parylene bi-layer samples in hot saline, which is a close imitator of body fluids. Serious delamination and water bubbles are found at the Parylene-Parylene interface (Figure 2.4(a)) of samples without any treatment, indicating that the Parylene sealing is not hermetic. It is also found, however, if the bi-layer films are treated with an annealing process for 3 days at 200 ºC in a vacuum oven (T-M Vacuum Products, Inc., Cinnaminson, NJ, USA), no interfacial delamination or water bubbles can be detected even after months of soaking (Figure 2.4(b)). This preliminary evidence reveals that high temperature annealing might be one of the approaches to improve Parylene sealing quality. In order to fully understand Parylene packaging behavior in saline, as well as to optimize the process for improving

28 Parylene sealing behavior, many experiments have been done which will be described in the following sections.

Figure 2.4: (a) Untreated Parylene bi-layer sample after one day soaking. Water bubbles are formed at the interface. (b) Annealed Parylene-bi-layer sample after long-term soaking. No delamination or water bubble is observed. 2.3 Interfacial adhesion enhanced process 2.3.1 Accelerated-lifetime soaking test of Parylene bi-layer structures A matrix of Parylene-bi-layer samples is designed and investigated under accelerated lifetime conditions to find reasonable process conditions for interfacial adhesion improvement. The samples are simply prepared by coating ~6 μm Parylene C on standard silicon wafers, followed by various processes on the Parylene surfaces. After that, another ~6 μm of Parylene C is deposited on top. Then bi-layer films are peeled from the wafers and cut into strips for test. Table 4.1 summarizes the preparation procedures of different samples, where oxygen plasma roughening is done in a PEII-A plasma etcher (Plasma Equipment Technical Services Inc., Livermore, CA, USA). This roughening process serves the purpose of increasing mechanical adhesion between the two Parylene

29 layers. In addition, ultrasonic and/or buffered HF cleaning are performed on some samples to remove contaminants and organic residues. Before testing, the samples are selectively annealed in the vacuum oven at setting temperatures ranging from 80 ºC to 260 ºC with different heating durations. The total chamber pressure is controlled to be approximately 10 Torr with nitrogen backfill, which is sufficient to prevent oxidative degradation of Parylene C at high temperatures (>120 ºC) in air. Table 2.1: Samples preparation of Parylene bi-layers for soaking tests. Treatments Sample set

Parylene deposition

1

6 μm

N/A

N/A

N/A

6 μm

2

6 μm

2 min

N/A

N/A

6 μm

3

6 μm

N/A

5 min

N/A

6 μm

4

6 μm

N/A

N/A

20 sec

6 μm

5

6 μm

2 min

5 min

N/A

6 μm

6

6 μm

2 min

N/A

20 sec

6 μm

7

6 μm

2 min

5 min

20 sec

6 μm

Parylene deposition Ultrasonic in water Buffer HF dip bath 200 mTorr, 400 W O2 plasma

The tested samples are held in centrifuge tubes filled with regular saline, and measurements are conducted in a convection oven at 77 ºC, as shown in Figure 2.5. During first month of soaking, the samples are removed daily from the saline for inspection using optical microscopy, and then at one-week intervals for the rest of soaking lifetime. To maintain a constant NaCl concentration, the test cells are cleaned and replenished once a

30 week. These tests have been continued for up to two years, and sample lifetimes are recorded when delamination or water bubbles are seen at the interfaces. Five samples are tested in each set, and their mean-time-to-failure (MTTF) is summarized in Table 2.2.

Figure 2.5: Testing cells of Parylene bi-layer samples are placed in a convection oven. Table 2.2: Accelerated-lifetime soaking test results of Parylene bi-layer samples. Red color denotes failed samples. Blue color denotes good samples. Annealing Treatments Sample set

80 ºC

140 ºC

180 ºC

200 ºC

260 ºC

12 12 12 12 12 2 days 2 days 2 days 2 days 1 hour 2 days hours hours hours hours hours 1

2 Y*

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2

1 D** 1 D

1D

2D

1D

1D

1D

1D

1D

5D

2Y

3

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

4

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

2Y

5

1D

1D

1D

1D

2D

2D

5D

7D

14 D 80 D

2Y

6

1D

1D

1D

2D

7D

40 D

21D

2Y

2Y

2Y

2Y

7

1D

1D

1D

6D

21 D

2Y

2Y

2Y

2Y

2Y

2Y

*: Year **: Day

31 From the test results, we find that O2 plasma roughening tends to exaggerate the delamination problem probably due to particle contaminants. Annealing at temperatures higher than 180 ºC can significantly improve chemical adhesion between Parylene and Parylene, even when devastating processes, such as O2 plasma roughening, are performed. As the temperature goes up, the process can convert poor-behaving samples into wellbehaving ones within a short time. It is of note that the annealing temperature has to be lower than the melting temperature of Parylene C (290 ºC), as strong recrystallization of Parylene C can happen beyond this point [50, 51], resulting in brittle Parylene C. Besides annealing, interface cleanness is also an important factor in order to ensure a good adhesion between two Parylene layers. According to the results of this test matrix, a routine procedure for optimizing Parylene adhesion is determined as follows: 1) Before subsequent Parylene deposition, buffered or diluted HF (10%) cleaning is necessary if microfabrications, such as O2 plasma roughening, photoresist processing, or chemical treatments, have been done at Parylene interface. 2) Post-fabrication annealing treatment needs to be performed in a vacuum oven for 2 days at 200 ºC. Mechanisms by which annealing treatment can improve Parylene to Parylene adhesion are still not completely understood. However, there are several hypotheses to explain it.

First, as can be seen from the results, annealing process takes effect at

temperatures beyond the glass-transition temperature of Parylene C (80-100 ºC) [52]. As a result, material reflowing could be accelerated at the interface, and polymer interweaving

32 between two Parylene layers could be increased. It is also possible that this process could provide sufficient thermal energy for activated ends of polymer strands at the interface to bond with each other. Finally, this annealing process might bake unpyrolyzed Parylene dimer out of the interfacial regions. All these effects, essentially, serve the purpose to convert the non-homogeneous two-layer structures into a more homogeneous bulk-like material. Consequently, water trapping at the interface could be reduced as most of water would pass through it instead [35]. However, it is hard to experimentally verify these hypotheses due to a lack of effective techniques to analyze interface properties. It is also well known that polymers tend to increase their density when heated and then re-cooled, especially if the excursion is beyond the glass transition temperature of the material. This fact suggests that the high temperature annealing might make Parylene denser so that water cannot permeate through thin films.

An experimental setup is

designed based on the water cup method to verify this hypothesis, and the experimental details and some of the preliminary results will be discussed in the next section. 2.3.2 Water vapor permeation through thin-film Parylene C Water vapor permeability is one of the important aspects to evaluate Parylene barrier properties, and particularly, water vapor transmission rate (WVTR) has been widely used as a measure for water vapor permeation of Parylene in the past. Table 4.1 gives reported WVTR values for Parylene C from literatures. However, all these data, which are measured from fresh Parylene C materials, are not applicable to annealed Parylene C. In addition, the water vapor permeability (WVP) of a polymer thin film depends greatly on

33 deposition quality and post-process conditions. Hence, studies need to be done in order to understand the permeability properties of our Parylene C. Table 2.3: Published WVTR data for Parylene C. (Courtesy of Parvathy Menon). Parylene C WVTR (g-mil/100in2-day)

WVTR at 37 ºC, 90%RH

Year of publication

Measured/company stated

Hubbell et al. [53]

3.3-7.3

1975

Measured

Loeb et al. [54]

0.5

1977

SCS [35]

0.21

2004

Company

SCS [55]

0.203

2007

Company

Caltech [56]

0.207

2008

Company (Union Carbide)

Company (Illinois Instruments)

A simplified water cup experiment has been designed to test Parylene WVTR in house. The experiment is performed in a small beaker filled with approximately 40 mL of DI water. The beaker opening is covered with a piece of Parylene C thin film and sealed with 5-minute dry epoxy. During the measurement, the beaker is stored in the cleanroom environment to maintain a constant relative humidity of 30% and a temperature of 20 ºC, and is weighed at distinct intervals using a Mettler analytic balance (Mettler Instrument Corp., Hightstown, NJ) to obtain its weight loss.

Then WVTR is calculated and

normalized to film thickness using the following equation [57]. WVTR =

MassLost × Thickness Time × Area

2-1

34 Table 2.4: Average WVTR of Parylene C thin films prepared under different conditions. Each sample set contains 2 samples from a same batch of Parylene deposition. (Data courtesy of Menon Parvathy[56].) Tested sample set

Annealing treatment

Avg. WVTR (gmil/100in2-day)

9 μm Parylene C

N/A

0.0638

9 μm Parylene C

200 ºC for 2 days

0.0387

18 μm Parylene C

N/A

0.0695

18 μm Parylene C,

200 ºC for 2 days

0.0435

18 μm bi-layer Parylene C

200 ºC for 2 days

0.1032

WVTR reduction 39.4% 37.4%

Various types and preparations of Parylene C are tested, and preliminary results are listed in Table 2.4. The results confirm that the annealing treatment indeed reduces the WVTR of Parylene C, and therefore improves Parylene water barrier properties by reducing perpendicular permeation through the thin films. A 9 μm film shows a WVTR reduction of ~40%, and similarly, an 18 μm Parylene sample shows a ~37% reduction. Comparing the bi-layer samples with the single layer samples, it is also found that the average WVTR of the bi-layer samples is approximately 60% higher, indicating Parylene to Parylene interface is not perfectly homogeneous even after annealing. This is the first experiment in which we are able to measure the WVTR of annealed Parylene, and the preliminary results show consistency with the soaking test results. Because WVTR is a function of system partial pressure and temperature, with a given WVTR under one condition, it is impossible to predict permeability properties under different conditions. Further measurements at various temperatures are underway. With

35 more data, we should be able to determine the activation energy of permeation, and moreover, to extrapolate WVTR reasonably well. 2.3.3 Parylene heat molding Another application of annealing treatment is Parylene heat molding.

For

biomedical applications, particularly, for intraocular implantations, Parylene devices, such as coil, electrode arrays and Parylene cables, needs to be modified into a curvature similar to that of the eye in order to meet surgical and comfort requirements. Instead of using prohibitively-difficult 3-D fabrication, Parylene C can be directly contoured and confined with an existing metal mold to achieve desired geometries. This heat molding technology will facilitate the device design and surgical procedures for human implantation, because different curvatures can be easily achieved by varying the size of molds to match the actual dimension of patients’ eyes.

Figure 2.6: Examples of Parylene heat molding: (left) Different shapes are formed for Parylene films. (right) Demonstration of flexibility after heat molding. Figure 2.6 (a) shows Parylene C films after heat molding at 200 ºC for 2 days. With the aid of metal molds, Parylene films can be formed into spherical or cylindrical shapes. In fact, it has been proven that this conformation is permanent due to Parylene glass transition, and therefore the conformation can be maintained throughout sterilization,

36 implantation, and follow up examination [35]. Additionally, as can be seem from Figure 2.6 (b), Parylene C still remains flexible after heat molding. We have also done heat molding for actual Parylene-based devices, as presented in Chapter 5. 2.4 Accelerated-lifetime soaking test of Parylene-metal-Parylen skins 2.4.1 Sample preparation Although we have demonstrated that the annealing process can significantly enhance Parylene to Parylene adhesion, no evidence shows that this process can be used to improve Parylene packaging behavior for embedded metals.

As Parylene-based

components in the retinal implant, such as RF coils and electrode arrays, normally involve Parylene-metal-Parylene sandwich structures, this is a crucial void in knowledge that must be addressed. To study the biostability of Parylene-protected metal, ALSTs have been done using a thin film resistor structure, which employs metal traces encapsulated by Parylene C. Samples are microfabricated in the same manner as described in Section 2.2.2 (Figure 2.7). Various samples are prepared with different metal combinations, Parylene thicknesses and post-fabrication treatments, as summarized in Table 2.5.

1 mm Figure 2.7: A Parylene packaged thin film resistor with pure gold metallization.

37 Table 2.5: Sample preparations. Sample set

Metallization

Parylen thickness (μm)

Annealing treatment

1

Ti-Au-Ti

~ 4.7

N/A

2

Cr-Au-Cr

~ 4.7

N/A

3

Pure Au

~ 4.7

N/A

4

Ti-Au-Ti

~ 4.7

2 days, 200 ºC

5

Cr-Au-Cr

~ 4.7

2 days, 200 ºC

6

Pure Au

~ 4.7

2 days, 200 ºC

7

Ti-Au-Ti

~ 9.2

2 days, 200 ºC

8

Cr-Au-Cr

~ 9.2

2 days, 200 ºC

9

Pure Au

~ 9.2

2 days, 200 ºC

2.4.2 Passive soaking test of Parylene-protected metal Samples are first tested passively in saline at temperatures of 77 ºC and 90 ºC respectively. To monitor their failures, the devices are removed from ovens daily, and observed optically under a microscope. Failure mechanisms of various samples are shown in Figure 2.8. Bubbles and delamination are first seen when failure is initiated in tested samples, mainly due to moisture permeation through Parylene C films or the ParyleneParylene interface. After that, metal corrosion is continuously propagated in the samples that contain adhesion promotion layers, namely titanium or chromium. The mechanisms of metal corrosion in body fluids have been studied by Mudali et al., and the results suggest that the corrosion is mainly caused by metal ionizations due to redox reactions or metal chloride breakdown [47]. Typical reduction potentials of Ti and Cr have been reported to be -1.630 V (Ti2+) and -0.913 V (Cr2+) respectively, which are much lower than that of Au

38 (1.498 V for Au3+) [58]. Therefore, Ti and Cr are easier to get oxidized when exposed to saline, which is consistent with our measurement results.

Figure 2.8: Microscope images of samples before and after passive ALSTs, showing different corrosion mechanisms. In each sample set, 5 samples are measured and their MTTF is recorded when the failure are seen (Table 2.6). With known MTTF at two temperature points, an Arrhenius relationship can be used to extrapolate the MTTF at body temperature [59]. In this model, the MTTF at a given system temperature T (in Kelvin) can be expressed by equation 2-2, where A is the pre-exponential constant, Ea is the activation energy of failures (in eV),

39 and k is the Boltzman’s constant (8.62×10−5 eV-K-1). The activation energy of each sample set is calculated using the Arrhenius relations, and their MTTF at body temperature is predicted, as summarized in Table 2.6.

MTTF = A exp(−

Ea ) kT

2-2

Table 2.6: Soaking test results for Parylene-metal-Parylene structures under passive conditions, and extrapolated data according to the Arrhenius relationship. Average MTTF in saline 90 ºC±2 ºC

Calculated activation energy Ea (eV)

Extrapolated MTTF at 37 ºC

90 minutes

35 minutes

-0.80

~ 45 hours

2

3 hours

80 minutes

-0.68

~ 55 hours

3

69 hours

44 hours

-0.38

~ 14 days

4

80 days

23 days

-1.05

~ 20 years

5

>250 days

28 days

---

---

6

>250 days

70 days

---

---

7

>250 days

>250 days

---

---

8

>250 days

>250 days

---

---

9

>250 days

>250 days

---

---

Sample set

77 ºC±2 ºC

1

The results show that the annealed samples from sets #5 and #6 have MTTF’s at body temperature well over 20 years, indicating the annealing treatment can significantly enhance the Parylene packaging performance by improving moisture barrier behaviors at Parylene-metal interface as well as at Parylene-Parylene interface. For the samples coated with 9.2 μm Parylene C, no major physical damage or water bubbles are discovered even after 250 days of soaking, confirming that thicker Parylene films will have better moisture

40 barrier performance than thinner ones.

Comparing the samples with various metal

combinations, we have found that Parylene-protected gold has the activation energy of approximately -0.38 eV (36.66 kJ/mol). This value agrees with the activation energy of water vapor permeation through bulk Parylene C [56, 60], indicating the failure is most likely caused by the moisture permeation in this case. However, the samples with adhesion promotion metal show lower activation energies due to metal oxygen reduction, which has free energies of activation normally ranging from -0.68 eV (66 kJ/mol) to -0.83 eV (80 kJ/mol) for many metals [61, 62]. 2.4.3 Active soaking test of Parylene-protected metal Active soaking tests are also performed using a simplified electrochemical setup (Figure 2.9), in which one electrode is the Parylene thin film resistor, while a platinum wire severs as a reference electrode. To connect a sample to test equipments, conductive wires are glued onto contact vias with conductive epoxy, and sealed with insulating epoxy to prevent unwanted current leakage. A square wave with a magnitude of 6 Volts and a frequency of 50 Hz is continuously applied to form a potential drop across two electrodes. The test cell is heated in a convection oven at 90 ºC, and the sample’s DC resistance is measured and recorded every 10 minutes using an HP 4145 B semiconductor parameter analyzer (HP/Agilent Technologies Inc., Santa Clara, CA) controlled by a LabView program (National Instruments, Austin, Texas). As we mentioned earlier, gold is the most favorable material for biomedical applications. Hereby, only sample sets with pure gold metallization (sample sets #3, #6 and #9) are tested under the active condition.

41

Figure 2.9: Simplified electrochemical test setup for active soaking test. A tested device is shown in the right picture. Figure 2.10 shows a representative curve of temporal DC resistance changes for samples from three sets. The sample failure is defined as time points when resistance is within 30% from the original value before, and increases significantly after, usually by over 3 orders of magnitude. Again, the annealed sample with thicker Parylene sealing shows a much longer life-time compared with others, which is consistent with the result from passive tests.

Figure 2.10: Typical aging curve of a sample: (top) from sample set #3. (middle) from sample set #6. (bottom) from sample set #9.

42 The failure mechanisms of samples under active soaking conditions are investigated using SEM and optical methods. Different failure modes are discovered in samples with different thicknesses of Parylene coating. The typical failure procedure on the samples with 4.7 μm Parylene is recorded at distinct day intervals using a Canon Digital Rebel XT camera (Canon Inc., Tokyo, Japan), as shown in Figure 2.11. Parylene breakdown is first seen after a short period of soaking, followed by Parylene delamination starting from the breakdown areas. Then metals are directly exposed to the saline and corrosion is observed. Finally, two Parylene layers are completely separated, and metal delaminates from Parylene.

Figure 2.11: Failure procedure observed in sample with 4.7 μm Parylene coating. The breakdown regions are also examined using scanning electron microscopy (SEM) and analyzed with energy dispersive X-ray spectroscopy (EDS), as shown in Figure 2.12. The EDS spectra confirm that the Parylene film is melted into a circular though-hole,

43 and the embedded metal is directly exposed. Parylene cracks radiate from the breakdown hole, which is probably caused by the thermal stress due to local heating. It should be noted that the breakdown hole is only observed on one side of the sample, while the other side remains intact, indicating the failure happens either on the interface or inside the Parylene film.

Figure 2.12: Typical SEM images and EDS spectra on the top side and the back side of the breakdown area.

44 Although the mechanism of the Parylene breakdown is not completely clear, we believe that it is defect-initiated breakdown because the breakdown is randomly located and localized in certain area where large defects and particles probably reside. Electrochemical reactions are accelerated at such points due to high electric field concentration and the direct exposure of metals to the saline, creating local metal corrosion. The breakdown hole has a diameter of over 6 μm (Figure 2.12 (a)), which implies that 4.7 μm of Parylene is not thick enough to cover some big defects or particles. Both annealed and unannealed samples show similar failure modes, indicating that the annealing process cannot help reduce defects or contaminant particles.

Figure 2.13: Even bubbles observed on samples from set #9 after soaking in saline at 90 ºC for 65 days. As for samples with 9.2 μm Parylene coating, only water bubbles are observed uniformly across the whole metal-Parylene interface (Figure 2.13), which suggests that moisture diffusion becomes the dominant failure mechanism instead of Parylene breakdown. It is possible that dimensions of defects or particles are relatively small under cleanroom conditions so that thicker Parylene could conformally cover them. If we use the

45 activation energy of water vapor permeation that we derived from passive conditions before, samples in set #9 will have an MTTF at 37 ºC of approximately 62 years. 2.5 Summary This chapter presented a Parylene-metal-Parylene skin technology, which is a key technology for making RF coils and multielectrode arrays in the retinal implant system. A standard procedure is also developed to improve the adhesion on Parylene-Parylene interface. To evaluate the long-term reliability of such Parylene-metal-Parylene skins, ALSTs are conducted in hot saline under various conditions. The results show that gold is less corrosive compared with titanium or chromium, and therefore is an ideal metal for biomedical applications.

It is also found that annealing at 200 ºC for 2 days can

significantly improve Parylene sealing behavior. In addition, Parylene film thicker than 10 μm is sufficient to cover small defects and particles produced from fabrication, preventing

electrical breakdown in the devices. The MTTF’s extrapolated using the Arrhenius model show that Parylene-protected gold can remain intact at body temperature for over 60 years, which is very encouraging for retinal implantation. However, there are still outstanding issues that deserve closer investigations. First, the ovens used for testing have temperature variation of +/- 2 °C. Due to the exponential nature of the Arrhenius model, this temperature change can affect the prediction significantly. Precise temperature control is needed to achieve a more accurate prediction. Also, because this soaking test is time consuming, we only analyzed data from a small sample set for initial investigation. Long term continuous tests are necessary to collect more data, which will further refine our knowledge on Parylene packaging behaviors.

46 C h a p t e r 3 DESIGN OF PARYLENE-BASED MEMS COILS

3.1 Introduction In Chapter 2, we have developed the basic technology for making flexible MEMS coils, involving a Parylene-metal-Parylene sandwich structure. We also know that metal lines can be protected by Parylene C from corrosive environments. However, there are other constraints on the MEMS coil design such as physical size and coupling efficiency, which are by no means straightforward to achieve. In general, MEMS coils suffer from low inductance and noticeable parasitic effect, namely parasitic resistance and capacitance, due to the characteristics of planar MEMS fabrication techniques. These effects are more pronounced with the intraocular coils because of their small physical dimensions. Therefore, it is critical to understand the coil electrical properties with respect to its geometric design, as well as the relationship between coil dimensions and the power transfer efficiency of the system, the essential of which will be covered in this chapter. 3.2 Inductive link system overview The all-intraocular retinal prosthesis system comprises an extraocular unit and an intraocular one. Power and data signal need to be transmitted from the extraocular unit to the intraocular unit for the stimulation of the retina and operation of all electronics. Our system relies on an inductive coupling link because it can provide the required power for a high-density retinal prosthesis and has relatively high efficiency compared with other power sources (Table 3.1). It also avoids permanent wired connections passing through the eyeball wall, minimizing undesired damages to eye tissues.

47 Table 3.1: Comparison of available power sources [63]. Battery Inductive

Infrared

Thermal

Transmitter

n/a

Coil

LED, Laser

Body heat

Receiver

n/a

Coil

Photodiode, phototransistor

Thermoelectric material

Available energy

Low

High

Moderate

Low

Efficiency

High

Moderate

Low

Low

Figure 3.1: System configuration for the proposed system [64]. Inductive coupling is an established field and extensive research has been performed to analyze and optimize the operation of inductive links for biomedical applications [65-68]. In our approach, dual-band architecture with two pairs of coils is proposed (Figure 3.1) [64], which uses different frequencies to transmit data and power independently. The operation frequencies are predetermined according to human tissue absorption and circuit design constraints. Specifically, the data signal is modulated on a 20

48 MHz carrier at a data rate up to 2 Mbps, while the power transfer is performed at a frequency between 1 to 2 MHz. The minimum required power transferred through this inductive link is estimated to be ~ 100 mW in order to enable high-density stimulations and regular chip functions [69]. The inductive link design also needs to comply with the safety limit of electromagnetic energy deposition into human tissue, as the intraocular unit will be operated in human eyes. One of the most prominent biological effects induced by RF radiation is local hyperthermia, caused by the thermal energy through absorption. The attenuation rate of RF radiation in tissues varies over the frequency spectrum. Studies about the average specific absorption rate (SAR) over the whole body indicate that maximum energy absorption takes place between 30 and 100 MHz for human beings (Figure 3.2). It is also found that human eye is particularly susceptible to heat. Microwave radiation at frequencies above 800 MHz can cause injury to eye [70]. The inductive link used in our approach operates at frequencies lower than 20 MHz, where little energy absorption is expected. The technical details of the inductive link circuitry are beyond the scope of this dissertation. The following sections will focus on the design procedures of MEMS receiver coils. A simplified inductive link model is also investigated to predict the power transfer capability of MEMS coils.

49

Figure 3.2: Average SAR for an average man exposed to 1mW/cm2 plane wave. (E polarization denotes the electric field parallel to the body, H denotes the magnetic field parallel to the body and K denotes the wave moving from head to toes or toes to head.) [70] 3.3 Design of planar MEMS coils In this section, we will concentrate on a circular shaped coil (Figure 3.3) to discuss the design procedure of a planar MEMS coil in general.

Figure 3.3: Simplified representation of a circular planar MEMS coil. 3.3.1 Self-inductance The self-inductance of a coil, Ls, is defined as the magnetic flux linkage per unit current in the coil itself. In 1945, Dwight arrived at some expressions for self-inductance

50 of circular coils with rectangular cross-section [71]. For a single layer coil with only one turn, the self–inductance can be calculated using the following equation:

⎡ ⎛ 4d ⎞ ⎛ ⎤ ⎞ 1 43t 2 t2 L0 = 2π d ×10−9 ⎢⎜ ln 1 + ⋅⋅⋅ ⋅⋅⋅⎥ (H), ⎟− + ⎟⎜ 2 2 ⎠ 2 288d ⎣⎝ t ⎠ ⎝ 24d ⎦

3-1

where d (in cm) is the mean diameter of the coil and t (in cm) is the coil width, as given in Figure 3.3. If the coil has N layers of metal wires and n turns in each layer, its inductance will increase significantly due to mutual inductance.

⎡⎛ 4 d ⎞ ⎛ ⎤ ⎞ 1 43t 2 t2 2 1 + ⋅⋅⋅ − + ⋅⋅⋅ Ls = 2π d (nN )2 ×10−9 ⎢⎜ ln ⎥ = (nN ) L0 (H). ⎜ ⎟ ⎟ 2 2 ⎠ 2 288d ⎣⎝ t ⎠ ⎝ 24d ⎦

3-2

It can be seen that the self-inductance is determined by the coil dimensions. However, this expression holds true only if the coil is operated at low-frequency, i.e., no skin-effect. 3.3.2 Effective series resistance The effective series resistance (ESR) of a planar MEMS coil, Rs, is used to estimate coil losses, which plays a significant role in designing a power efficient inductive link. In general, the ESR can be divided into two parts; one part is DC resistance and the other one is frequency dependent. For a single turn coil, its DC resistance can be calculated using Ohm’s law, where ρ is the resistivity of the conductive material and h is the thickness of the metal wire.

51

R0 = ρ

πd t×h

(Ω).

3-3

Assuming the width of metal traces is much larger than the space between adjacent turns which can be ignored, then the DC resistance of a coil with N layers of metal and n turns on each layer can be expressed as Equation 3-4.

Rs = ρ

nN × π d = n2 NR0 (Ω). t ×h n

3-4

The frequency dependent part, on the other hand, is caused by time varying magnetic fields produced by an alternating current passing through the conductor. This field can introduce eddy currents, which flow in the opposite direction to the applied currents. As a result, the current tends to shift to the surface of the wire, leading to an increase in the series resistance. This phenomenon is well known as “skin-effect”, and the skin depth of a given conductor can be calculated by the following equation[72, 73], in which μ0 is the permeability of free space ( 4π ⋅ 10 − 7 H / m ), μr is the relative permeability of conductive material, and ω is the angular frequency.

δ=

2ρ

ωμ0 μr

(m).

3-5

In our retinal implants, we use gold wires ( ρ ~ 2.2 ×10−8 Ω ⋅ m and μr ~1 ) at frequencies ranging from 1 to 2 MHz. Therefore, the skin depth δ is calculated to be approximately 53 μm to 75 μm, which is by far greater than typical metal thicknesses deposited in an E-beam evaporator. Therefore the influence of the skin-effect is negligible and the coil ESR can be expressed using its DC resistance only (Equation 3-4).

52 3.3.3 Parasitic capacitance Parasitic capacitance places a limit to the coil self resonant frequency, above which the coil will not behave as an inductor any more. The parasitic capacitance of a planar MEMS coil usually consists of three different capacitances: the capacitance between coil turns, the capacitance between layers, and the capacitance between the coil and substrate. In our case, silicon substrates will be completely removed, therefore only capacitances between turns and layers are considered. A distributed model is used to approximately estimate the equivalent parasitic capacitance [74]. In order to simplify this model, several assumptions are made: 1) The spacing between adjacent turns is much less than the width of metal traces, therefore it can be ignored. 2) The metal width and thickness are consistent everywhere so that voltage distribution is proportional to the lengths of metal traces. 3) Within the same turn, voltage is equally distributed, which is determined by averaging the beginning voltage and the ending voltage. 4) Only capacitances between adjacent turns and layers are taken into account. The capacitances between non-adjacent turns and layers are second order effects and will be neglected in this simplified model.

53

Figure 3.4: Voltage profile of an n-turn planar coil. [74] Figure 3.4 shows voltage profile of a single layer coil with n-turn metal traces. The voltage in k-th turn can be express as Equation 3-6, where d (k ) ≡ h1 + h2 + ... + hk , and hk is defined as the ratio of the wire length of the k-th turn (lk ) to the total wire length (ltot).

1 1 V (k ) = [V (k )beg + V (k ) end ] = V0 [2 − d (k − 1) − d (k )] . 2 2

3-6

Hence, the voltage drop between k-th and (k+1)-th can be derived as:

1 ΔV (k , k + 1) = V (k ) − V (k + 1) = V0 [d (k + 1) − d (k − 1)] . 2

3-7

The total reactive energy stored in the distributed capacitors can be expressed as: n −1 n −1 1 1 1 Etotal = ∑ C (k ) ⋅ ΔV (k , k + 1)2 = ∑ Cmmlk ⋅ {V0 [d (k + 1) − d (k −1)]}2 , 4 k =1 2 k =1 2

3-8

where Cmm represents the capacitance per unit length between adjacent metal turns. Based on the fact that this total reactive energy is equal to the energy stored in the equivalent lump capacitor, we can obtain the equivalent capacitance between adjacent turns by

54 n −1 1 Ceq−t = ∑ Cmmlk [d (k + 1) − d (k − 1)]2 . k =1 4

3-9

A similar distributed model is applied to an N-layer stacked coil, where the voltage profile across the stacked layers is shown in Figure 3.5. The equivalent capacitance between adjacent layers is given as Equation 3-10:

Figure 3.5: Voltage distribution of an N-layer coil. A 1 N (Cm ,m −1 + Cm − 2, m −3 + ...) k2 [4 − 2d (k − 1) − 2d (k )]2 ∑ 4 k =1 m , N Ak 1 2 + ∑ (Cm −1, m − 2 + Cm −3,m − 4 + ...) 2 [2d (k − 1) + 2d (k )] 4 k =1 m

Ceq −l =

3-10

55 where Cm, m-1 is the capacitance per unit area between the m-th and (m-1)-th metal layer, and Ak is the trace occupied area of the k-th turn. Therefore, the total parasitic lump capacitor is the sum of Ceq −t and Ceq −l . 3.3.4 Self-resonant frequency and Quality factor The self-resonant frequency and quality factor (Q factor) of coils can be derived using an equivalent RLC model (Figure 3.6). The self-resonant frequency of this RLC tank is calculated as ωs =

Rs2 1 Ls 1 − 2 . When Rs 1,000) integration.

Finally, the

favorable mechanical and chemical properties of Parylene C (Section 1.3.2 ) allow our final integrated system to be highly flexible and biocompatible for implantation.

5.2 Accelerated-lifetime soaking test of Parylene protected ICs Before delving into the integration technology, it is important to study the Parylene sealing performance on IC circuitry.

To do that, similar ALSTs are performed on

Parylene-protected IC chips. An EM4100 read-only radio-frequency identity (RFID) chip (EM Microelectronic, Switzerland) is used as a sample chip. This chip is a CMOS integrated circuit with metal composition of Al-Ni-Au on contact pads. It can be powered up using an external coil with an electromagnetic field. By turning the modulation current on and off, the chip sends back a 64-bit sequence stored in the memory array. The operation frequency of the chip is between 100 kHz and 150 kHz, with 125 kHz being typical. Unpowered soaking test is first performed in saline at 77 ºC. Samples are prepared by cleaning chips with isopropanol and DI water, followed by a Parylene C deposition of approximately 10 μm. Bare chips without Parylene coating are also tested as comparisons. During the soaking period, the samples are examined daily under a microscope. The test lasts up to six months, and then the samples are removed from hot saline for functionality measurement using a Wentworth Labs 11PO900 probe station (Wentworth Laboratories, Inc., Brookfield, CT, USA). As can be seen from Figure 5.1, the chip without Parylene coating experiences serious metal corrosion after only 2 days of soaking, while the samples

91 with Parylene protection has not observable metal damage even after six months of soaking. Functionality test is performed on the Parylene coated chip after removing the Parylene on contact pads with oxygen plasma etching or probe tip scratching. The result shows that the chip is still well functioning, indicating that IC circuits can remain intact in hot saline for six months with Parylene protection.

Figure 5.1: Microscope images of chip metal pads. (a) A chip coated with annealed Parylene C. (b) A chip with regular Parylene C after 6 months of soaking. (c) A bare chip after 2 days of soaking. Given this exciting result, active soaking tests are conducted to further investigate the lifetime of the Parylene protected chip under continuous electrical stressing. In this case, the RFID chip is glued onto a Parylene cable using a biocompatible silver epoxy EPO-TEK H20E (Epoxy Technology, Billerica, MA, USA). The epoxy is fully cured and outgassed in a convection oven for 3 hours at 80 ºC. Regular wires are also pasted on the other side of the Parylene cable in order to connect the chip to external test units, as shown in Figure 5.2. The device is then conformally coated with Parylene C for testing. The soaking test is performed in a convection oven. During the test, the device is held in a glass bottle filled with regular saline, and the wires are connected to an external coil. The external hand-wound coil has an inductance of ~5.3 mH and an ESR of 88 Ω at 125 kHz. A commercially available RFID reader module (Parallax Inc. Rocklin, CA USA)

92 is used to continuously send power and requests (24 hours a day) to the RFID chip though inductive coupling. The reader module contains an integrated coil with dimensions of 60 mm by 66mm, and is driven by a 5 volt DC supply. The output port of the reader is connected to an HP 54503A oscilloscope and a webcam is used to monitor the signal readout by taking a picture every 10 minutes (Figure 5.3).

Regular wires

Parylene cable

Figure 5.2: (left) Assembled RFID chip soaked in saline. (right) Close-up image of the attached RFID chip. RFID reader

Hand-wound coil

Figure 5.3: Test setup for the active soaking test of RFID chips.

93 Samples coated with different thicknesses of Parylene are tested at various temperatures. MTTF is defined as the time point when no output signal can be detected. Four samples are tested under each condition, and their MTTF and the standard deviation is plotted in Figure 5.4.

According to the Arrhenius relationship, chips with 22.1 μm

Parylene coating will have an MTTF of ~ 2 years at 37°C under the active condition. Ideally, samples with thicker Parylene coating should have better durability if failure is solely caused by water permeation through Parylene. However, the results show that samples with 22.1 μm Parylene coating do not show much advantage, indicating that other failure mechanisms besides water permeation might have taken place. 25

MTTF (Days)

20

Standard deviation Average MTTF

15 10 5 0 22.1 µm at 77°C 22.1 µm at 85°C 11.7 µm at 85°C Samples

Figure 5.4: Active soaking test results of Parylene protected IC chips. To find out the exact failure mode, the failed samples are examined using SEM. Parylene cracks are observed in samples with both 22.1 μm and 11.7 μm Parylene C coatings. The mechanism of Parylene cracking has not been fully understood. It is possible that the chip generates a lot of heat under intensive electrical stress so that thermal stress concentrates along the edges due to poor heat dissipation. Also, it is noted that the

94 test temperatures are very close to or within the range of the glass transition temperature of Parylene C (80 °C – 100 °C), therefore Parylene might undergo material property changes due to the recrystallization of the polymer chains. The active soaking test is still ongoing, and more experimental results are expected to enable a complete understanding of Parylene packaging behaviors.

Figure 5.5: Parylene cracks is sample with (left) 11.7 μm Parylene coating and (right) 22.1 μm Parylene coating.

5.3 Integration demonstration 5.3.1 Design and fabrication Figure 5.6 depicts a conceptual schematic of an embedded chip integration, in which individual prefabricated chips and passive components can be embedded in a carrier silicon wafer and sealed with Parylene C. After that, the fabrication of other surface MEMS components, such as electrode arrays and coils, can be performed on the same platform. To demonstrate this technology, the same EM4100 read-only RFID chip is used again as shown in Figure 5.7 (a). In order to determine the chip housing dimensions, the chip is imaged with a WYKO interferometer (Veeco Instrument Inc., Woodbury, NY,

95 USA), and is measured to have average dimensions of 1 mm in length, 0.98 mm in width, and 182 μm in thickness (Figure 5.7 (right)). A square planar coil is designed for wireless power and data transmission, which consists of two metal layers with 22 turns on each layer. The overall coil size is 2 cm by 2 cm. During the fabrication, the two coil terminals are connected to Pad_1 and Pad_2 on the chip (Figure 5.7 (left)). Due to the low power consumption of the wireless chip, only the coil is needed to power the circuit, and no extra capacitor is required.

Figure 5.6: Concept of embedded chip integration.

Figure 5.7: (left) EM 4100 RFID chip used to demonstrate the integration technology. (right) WYKO image of a typical chip. Figure 5.8 illustrates the detailed fabrication process, starting with a standard 4 inch wafer that is ~550 μm thick and coated with a layer of sacrificial photoresist. A 5 μm layer

96 of Parylene C is deposited on top of the photoresist. To secure Parylene on the substrate, Parylene anchors surrounding the chip housings are etched into the substrate using Bosch process in a DRIE system (Unaxis Corporation, St. Petersburg, FL, USA) [87]. After Parylene deposition, a 500 nm layer of gold is deposited using an E-beam evaporator, and patterned to form the first layer of coil wires. Next, cavities matching the chip dimensions are etched into the Parylene and silicon using RIE and Bosch process, and the chips are then dropped into the cavities to self-align with an average lateral displacement of less than 10 μm. During this procedure, epoxy is applied to the bottom of the cavities as needed in order to compensate for cavity depth inaccuracy and to fill the gaps surrounding the chips. A metal pole with a flat silicon piece attached on one end is used to push the chips into the cavities and to level the surface (Figure 5.9). After the chips are fitted into the cavities, Parylene deposition is performed again to seal the chips and to form an insulation layer between two metal layers, followed by oxygen plasma etching to properly open interconnection vias. A second 500 nm layer of gold is then deposited and patterned to form the top layer of coil wires and interconnects. Another 5 μm layer of Parylene C is coated on top and patterned using oxygen plasma to define the profile of devices. Through-wafer trenches surrounding the chips are then etched from the backside using the Bosch process, and the Parylene-based coil is released from the substrate, carrying the chips encapsulated in silicon housings. Parylene for protection.

Lastly, the entire device can be sealed with

97

Figure 5.8: Detailed process flow for embedded chip integration.

Figure 5.9: Metal pole with a silicon piece for use to push the chip to the silicon shell. In this integration technology, minimizing the vertical displacement of the chip from the wafer surface is crucial for subsequent photolithography and metallization steps. Large vertical displacements will result in broken wires around the chip. To better control the process, the surface heights of the chips are measured using a stylus profilometer P-15 (KLA-Tencor, San Jose, CA, USA) after the chips anchor in place in the silicon shell.

98 Figure 5.10 shows the surface profiles of four successfully integrated chips with respect to the surrounding wafer. The overall vertical displacements from the wafer surface are within 6 μm, providing sufficient flatness for further processes.

Figure 5.10: Vertical displacements of chips: 0μm step height indicates wafer level. Another important aspect is the horizontal alignment of perimeter interconnects to the embedded chips, which limits the interconnection density of the integration. By design, the chip should be self-aligned in the silicon shell within 10 μm of lateral displacement; however some chips are misaligned by more than 10 μm due to the dimension variation of the chips themselves (Figure 5.11). With tighter tolerance on cavity sidewalls, or with more precise chip-alignment lithography, this alignment deviation can be reduced in the future to achieve high-lead-count integration.

99

Figure 5.11: (a) Example of 10 μm lateral misalignment of the chip. 5.3.2 Fabrication and functional test results We have successfully integrated the RFID chip with the Parylene-based RF coil, and Figure 5.12 (a) shows an example of fabricated devices. In this device, the selfinductance of the receiver coil is around 16 μH measured with an HP 4192 impedance analyzer, and the DC resistance is around 275 Ω measured with an Agilent 34401 multimeter. The total thickness of the Parylene-metal-Parylene thin film is approximately 20 μm. Thank to the flexibility of Parylene, this final device is very flexible, allowing the device to be placed in direct contact with delicate tissues during surgical implantation (Figure 5.13).

100

Figure 5.12: A fabricated device: (a) Overall view of the device. (b) Close-up view of the embedded chip. (c) Close-up view of the coil wires.

Figure 5.13: Flexibility of the Parylene-metal-Parylene thin film. The functionality of the embedded RFID chip is tested with the Parallax RFID reader module, as shown in Figure 5.14. The output port of the reader is connected to an oscilloscope to monitor the signal readout. When the coil-integrated RFID is placed within reading distance of the reader, the information stored in the chip will be detected by the reader, and an 8-bit transistor-transistor logic (TTL) signal sequence can be observed on the oscilloscope, as shown in Figure 5.15. The spatial separation between the primary coil and secondary coil is varied during measurement, and a maximal detectable range of 3 mm to 4

101 mm is found. This short reading distance is mainly due to the low Q-factor of the receiver coil which is calculated to be approximately 0.05 at 125 kHz, and therefore results in low power transfer efficiency through the electromagnetic field.

Figure 5.14: Test setup to verify the function of embedded RFID chips.

Figure 5.15: Signal readout from the oscilloscope. 5.4 Mechanical model for implantation studies We have developed and demonstrated an integration technology to incorporate a prefabricated chip in a Parylene-based system. Naturally, the next step is to build a monolithic intraocular system using this integration technology, followed by the

102 demonstration of in vivo retinal stimulation using such a system. However, due to the small volume and complex anatomy of human eye, it is important to carefully evaluate possible system geometries prior to finalizing the specifications. In addition, it is necessary to study the feasibility of surgical implantation in animal models to determine optimal implantation techniques, and to verify the robustness of the devices and materials under surgical manipulation. Therefore, an intraocular test system is designed (Figure 5.16) and fabricated in Parylene C using a photoresist mask and oxygen plasma etching, simulating the mechanical integration of various key components using size-matching dummies. The design includes an RF coil, intended to be implanted in front of the iris. A flexible cable is passed through the anterior and posterior lens capsule into the vitreous cavity, and an electrode array with a 450 μm diameter tack hole is attached to the retina using a retinal tack. In order to reduce fabrication complexity, the coil part and the array part are separately fabricated in this dummy system. The coil cable is heat-molded to form a “Ushape” curvature after fabrication, rounding the anterior and posterior capsulotomies into the vitreous cavity during implantation. Dummy coils, which have the same dimensions as final devices, are then glued onto the coil part using biocompatible epoxy EPO-TEK 301-2. Similarly, the array part is heat shaped using a custom 6061 aluminum mold comprising a recessed concave region and a mating stainless steel sphere that approximates the curvature of the canine retina (diameter ~22 mm), as shown in Figure 5.17. During annealing, the array part is sandwiched between the sphere and the mating surface, while the cable is pressed flat against the aluminum surface. After that, dummy chips are glued onto the Parylene cables in one of the possible positions anticipated for the ASICs. Finally, the two

103 separate parts (the coil region and the array region) are bonded together using epoxy. The curing condition for this particular epoxy is 80 °C for 3 hours in a convection oven.

Tack hole

Figure 5.16: Prototype geometry for an all-intraocular Parylene-based device with all required component regions for a completely implantable system.

Figure 5.17: Custom heat-forming mold. (Images courtesy of Dr. Damien Rodger) Two prototypes for acute implantations are designed and fabricated (Figure 5.18), showing an overall Parylene thickness of approximately 18 μm. For the model designed for test chip integration, the coil is made from aluminum foils with dimensions of ~3.75 mm in inner diameter, ~9 mm in outer diameter, and ~200 μm in metal thickness. The test chip used in this model is a BION stimulator designed for neuromuscular prostheses [88]. As for the IRP model, the coil and the chip model are both made of acrylonitrile/butadiene/styrene (ABS, RedEyeRPM, Eden Prairie, MN, USA). The coil has dimensions of ~7 mm in inner diameter, ~11 mm in outer diameter, and ~1.5 mm in thickness. The chip model is a hexahedron with dimensions of 6 mm × 6 mm × 2 mm. In

104 the final system, this part is actually a hexahedral can which has all electronic components integrated and packaged in it.

Figure 5.18: Two surgical models for acute implantation: model-I (a) is designed for test chip integration; model-II (b) is designed for IRP1K integration. These test models are then implanted in a canine eye to evaluate surgical approaches and to determine the system geometrical feasibility inside the eye. Example surgical photographs are given in Figure 5.19. A significant problem with these initial designs is that the chip positions are not optimal, which induces surgical difficulties. For instance, the chip is too close to the coil site in the test chip surgical model, blocking the view during implantation so that it is difficult to see the tack hole on the array site. In the IRP surgical model, the dummy chip is attached partway down the Parylene cable, therefore the view is less obstructed. However, we still have trouble with locating the tack hole because the chip module is near the array part and blocks the view of the surgeon during the operation. In addition, the coils are floating in the anterior chamber in both cases, but they do need to be anchored into the angle to achieve excellent alignment and coupling. Finally, the cable length is too short for this canine eye, which tends to draw the coil against the iris during array tacking. This could become a serious problem if the pulling force was too big, which would damage delicate tissue in eye.

105

Figure 5.19: Example surgical photographs of prototype geometry implantation using (a) test chip surgical model; and (b) IRP1K surgical model. To address the surgical problems encountered by the first batch of prototypes, an improved model is designed (Figure 5.20) in which the chip part is moved further away from the coil part, towards the middle part of the Parylene cable. The chip is attached along the cable rather than perpendicular to the cable. The inner diameter of the coil is increased to 5 mm, allowing a clearer view of the fundus during and after the surgery. To anchor the coil to the angle of the anterior chamber, a commercially available anterior chamber intraocular lens (ACIOL) (Model # 122UV, Bausch & Lomb, Rochester, NY, USA) is half-cut, and the haptics are attached to the coil, as show in Figure 5.21. After all components are glued together, a 1 μm layer of Parylene is deposited everywhere to

106 improve the biocompatibility of final devices. Two cable lengths are investigated for canine eyes, while the precise dimensions will vary for human implantations.

Figure 5.20: Improved geometry for a completely implantable system.

Figure 5.21: Example of a coil with ACIOL haptics attached. The proposed devices are fabricated (Figure 5.22), and tested for in vivo implantation is operated at the Keck School of Medicine of USC. First, an incision approximately 10 to 11 mm long is cut in the cornea, followed by an anterior capsulorhexis. The lens is then extracted using phacoemulsification. After that, a posterior capsulotomy is made, and the vitreous is removed from the posterior chamber. The chip and array parts are then inserted into the posterior chamber through both holes in the lens capsule, leaving the coil part anchored in front of the iris. Vitrectomy is performed through the sclera ports, and the array part is tacked in place on the retina. Finally, the cornea and

107 the sclera ports are sutured closed, and the dogs are kept alive with devices implanted to evaluate the mechanical and biocompatible stability of the devices inside the eyes. Figure 5.23 gives examples of intraoperative surgical photographs.

Figure 5.22: Two surgical models for chronic implantation in canine eyes.

Figure 5.23: Surgical implantation photographs of improved surgical model under: (left) anterior illumination, and (right) intravitreal illumination. The surgical implantation indicates that the modified design results in a better surgical outcome, and that the haptics significantly help the coil region to anchor within the angles. However, serious inflammation is observed in both eyes a week after implantation, which is probably caused by toxins released from the epoxy. To further address this matter, the epoxy needs to fully cure and outgas before implantation. A thicker Parylene coating on the epoxy will also improve the isolation of the epoxy from eye environment.

108

5.5 Test chip system integration 5.5.1 Functional system prototype A completely integrated system is designed according to the preliminary results of the surgical implantation of the mechanical models, employing the BION chip as a test chip. The BION chip is a single-channel stimulator chip which is initially developed to mimic muscle spindle function, and to help patients who are suffering from muscle paralysis. Once implanted into paralyzed muscles, the chip can receive both power and command signals through inductive coupling over a 480 kHz~500 kHz power carrier generated in an external unit, allowing them to emit precisely timed stimulation pulses with a highly regulated amplitude and pulse-width [88]. There are several reasons to select the BION chip as the test chip. First of all, chips specifically designed for retinal implantation are currently unavailable. The BION chip therefore becomes a reasonable substitute since a BIONic system includes all major components required for a final retinal implantation system. In addition, the BION chip consumes less power compared with the circuitry used in retinal implantation so that our current coil prototype might be able to deliver sufficient power for the chip operation.

Finally, its integration process will involve the same

embedded chip technique which can be easily repeated for retinal implantation. Figure 5.24 shows the layout of the BION chip, and the pad functionalities are described in the right table. This chip has small physical dimensions of ~1 mm in width, ~2.33 mm in length, and ~257 μm in thickness. There are five pads on the chip with dimensions of approximately 120 μm × 74 μm. The distance between adjacent pads is approximately 110 μm.

109

Figure 5.24: BION chip layout and the pad connection. The integrated system consists of a BION1-3 CMOS chip, two capacitors, a specially designed MEMS coil for RF power and data transmission from an external personal trainer, as well as a carrier substrate with a 2-electrode array, interconnection leads, and contact pads for system assembly. Physical dimensions of the devices are to meet the specifications predetermined by previous surgical implantation results. Figure 5.25 shows a system schematic and its corresponding circuit connections, where the onchip rectifier is represented using a full-wave rectifier. C2 is an off-chip charge storage capacitor which has a capacitance of ~22 nF, provided by Loeb’s group. C1 is a tuning capacitor in parallel with the receiving coil in order to achieve a resonant frequency of approximately 480 kHz. C3 is used to simulate the tissue capacitance between the ground electrode and simulation electrode, which varies in different implantation environments, and thus is not included in our integration system.

Ceramic chip capacitors (AVX

Corporation, Myrtle Beach, SC, USA) are selected because they are about 1 mm × 0.5 mm × 0.56 mm in size and can be embedded in a standard silicon substrate.

110

Figure 5.25: Schematic overview of a BION chip integration system. 5.5.2 Fabrication of the assembled BION system Before building a monolithic system with the complicated CL-I2 integration technology described earlier, it is important to calibrate each individual component, as well as to verify the system functionality. Therefore, it is worthwhile to make an assembled system by interconnecting components together on a Parylene substrate with conductive silver paste. The fabrication is divided into three steps: carrier substrate fabrication, RF MEMS coil fabrication, final system assembly and packaging. Figure 5.26 depicts the detailed process for making the carrier substrate. Parylene-C is first deposited on a silicon substrate. Then a 200 nm layer of metal is E-beam evaporated on the Parylene, and

111 patterned using lift-off to form connection pads, interconnection leads, and electrode sites [35]. Platinum is selected as the electrode material for optimal simulation capability, and no adhesion metal layer is needed in this case because platinum and Parylene are known to have good adhesion. After metal lift-off, another layer of Parylene is coated to seal the entire structure, followed by oxygen plasma etching in an RIE machine with a photoresist mask to define the outer geometry of the substrate, as well as to open the electrode sites and the contact vias. Finally, the device is peeled off from the silicon substrate in a water bath.

Figure 5.26: Fabrication process for the Parylene-based carrier substrate. Steps (b)-(d) describe the lift-off technology for platinum patterning. Figure 5.27 presents a fabricated device, and the microscope image shows a special chip site design, where Parylene ribbons are etched in a way such that the chip can be held in place and aligned to the contact vias on the carrier substrate during assembly.

112

Tack hole

Figure 5.27: (left) Fabricated carrier substrate. (right) Parylene ribbons to hold the chip.

Figure 5.28: A fabricated fold-and-bond coil with two layers of metal. The electrical characteristics are measured and given in the table. To fabricate the MEMS coil, the fold-and-bond technology is involved as discussed in Section 4.3.2 . Two single coils are placed in series, and made of a 3 μm layer of gold to achieve low resistance and high Q-factor. Then the device is folded into two layers and bonded together thermally. A fabricated fold-and-bond coil is shown in Figure 5.28, comprising two layers of metal with 10 turns in each layer. In-and-out leads are connected to the carrier substrate from the center, facilitating the surgical procedure. Through vias are designed to overlap with the contact pads, so the interconnections to the carrier substrate can be formed from both sides. Coil electrical properties are measured, showing an

113 inductance of ~ 2.24 μH and a DC resistance of ~ 15.82 Ω, which gives a Q factor of 0.45 at 500 kHz. For hybrid system assembly, individual components (the BION chip, the coil and the capacitors) are aligned to the corresponding interconnection vias on the carrier substrate. A small amount of biocompatible silver epoxy (EPO-TEK H20E) is then applied on the contacts and cured at 80 ºC for 3 hours. The conductive epoxy serves two purposes: to form the interconnections between the components as well as to bond the components onto the carrier substrate. Figure 5.29 shows an assembled BION system, and close-up views on the interconnections for each individual component.

(a)

(b) Figure 5.29: (a) A assembled BION system. (b) Interconnects formed with biocompatible conductive silver epoxy: (left) coil contacts, (right) chip and capacitor contacts.

114 Although this epoxy interconnection method is quick and easy, a main problem encountered during the fabrication is the lack of control of the epoxy amount when applied by hand (Figure 5.30). Short circuits can be created if too much epoxy is applied. Epoxy reflow can also happen during high temperature curing, resulting in short circuits. For chips with high density pad layouts, hand assembly is no longer applicable, and thus wafer level integration technology is necessary.

Figure 5.30: Examples of chip interconnections. For the right sample, short circuit is formed between adjacent pads. Another big concern for this assembly technology is the isolation of the silver epoxy from the eye environment. It is well known that human eye is a delicate and sensitive organ, and any toxin released from the epoxy can cause inflammation and infection. In addition, eye fluids are very corrosive to exposed metal, which can reduce the durability of silver epoxy bonding, and finally result in open circuits. In order to overcome these problems, a layer of Parylene is coated to seal the entire system after assembly, just leaving the electrode site open.

115 5.5.3 Functional test of the assembled BION system After the assembled system is made, its functionality is verified using a telemetry setup, as shown in Figure 5.31. The primary stage, which comprises a personal trainer unit, a class-E coil driver and a hand-wound transmitting coil, can generate a power carrier of ~500 kHz. The personal trainer stores command programs personalized for individual subjects, records the time and duration of treatment, and transfers this information to a computer for real time monitoring. Up to three programs can be loaded into the memory of the personal trainer [89]. The coil driver is connected to the personal trainer through a custom made adapter. The transmitting coil has an inductance of ~ 46.4 μH and a Q factor of ~118 at 500 kHz. Litz wires 1025-44 SPN are used for winding the primary coil in order to reduce the skin effect and proximity effect losses, where 1025 represents the number of strand in each bundle, and 44 represents American wire gauge per strand which is 50.8 μm by definition. The transmitting coil is built in a solenoid shape to establish a more uniform electromagnetic field inside the coil. Additionally, ferrite cores are incorporated in the transmitting coil to magnify the electromagnetic field and to improve voltage transfer efficiency, based on the fact that the induced voltage at the receiving coil increases proportionally with the increase of magnetic flux going though the inductive link.

116

Figure 5.31: A telemetry setup for functionality test of the assembled BION system. The right picture shows a personal trainer and other peripheral accessories. In vitro measurements have been conducted, and output signals are monitored by connecting two electrodes directly to an HP54645A oscilloscope. Figure 5.32 shows a stimulation pulse recorded from the electrode sites with a pulse width of approximately 500 μs and an amplitude of approximately 7 V. In order to estimate the power transfer capability of the MEMS coil, the voltage across the receiving coil terminals is measured. The current delivered to the chip is also obtained by measuring voltage on an extra resistor in series with the chip. Typical waveforms of the transferred voltage and current are given in Figure 5.33, showing a resonant frequency of ~505 kHz and a ~25 degree phase drift between voltage and current. As mentioned earlier, the tuning capacitor is a commercial chip capacitor, which has limited options of capacitance values. Therefore, it is difficult to fine tune the resonant circuit to achieve precise synchronization.

117

Figure 5.32: Typical simulation pulse measured from the electrode sites.

Figure 5.33: Waveforms of transferred voltage and current to the BION circuitry. The distance between the two coils is varied, and stimulating pulses, transferred voltage as well as delivered power at the different separation distances are recorded and plotted in Figure 5.34. The current system has a detectable range of ~ 4 mm, which is mainly limited by the low Q factor of the receiving coil.

The output pulses have

amplitudes ranging from ~6.8 V to 8.3 V, depending on the separation distance between the coil pairs. The figure also shows a threshold for the transferred voltage near 7 V, below that no stimulating pulse can be detected from the output electrodes. The maximum power delivered through the inductive link is around 43 mW, and drops dramatically by ~ 62% as

118 the separation distance increases to 2 mm due to the divergence of the electromagnetic flux at the edge of the primary coil. (b) Tansferred voltage

9 8.5 8 7.5 7 6.5 6

Vmax (V)

Pulse amplitude (V)

(a) Pulse amplitude

0

2

4

12 11 10 9 8 7 6

6

0

Separation distance (mm)

2

4

6

Separation distance (mm)

Transferred power (mW)

(c) Tansferred power 50 40 30 20 10 0 0

2

4

6

Separation distance (mm)

Figure 5.34: Measured stimulating pulses (a), transferred voltages (b) and transferred power (c) at different separation distances between two coils. 5.5.4 Monolithic system processing After demonstrating the functionality of the assemble system, a completely monolithic system is designed and currently undergoing process characterization and refinement, using the BION as the test chip. The complete fabrication process flow is given in Figure 5.35, where the RF coil, platinum 2-electrode array, BION chip and capacitors are integrated in the same platform.

119

Figure 5.35: Integration process for monolithic neural stimulator. Because the coil leads are connected to the carrier substrate from the center, a crossover bridge is formed on the overlap area by inserting a layer of sacrificial photoresist and removing it afterwards. One issue associated with this change is that cracks tend to form in the photoresist, and delamination could occur between Parylene and silicon

120 substrate (Figure 5.36). Some ongoing efforts aim to address these problems and optimize the fabrication processes.

Figure 5.36: Photoresist cracking and Parylene delamination encountered during the monolithic integration processing. Nevertheless, we have demonstrated our ability to integrate IC circuitry with MEMS components for a fully functional system using the RFID chip. A functional neural stimulator featuring the BION chip has also been implemented. Although a monolithic integrated device has not been fabricated yet, we believe it is just a matter of time. Future work should include finishing the monolithic integration process, and implant the integrated system in animal model to verify its functionality in vivo.

5.6 Summary In this chapter, an embedded chip integration technology has been developed and successfully demonstrated by integrating a flexible Parylene RF coil with a commercially available read-only RFID chip. The functionality of the embedded chip is tested using an RFID reader module. Preliminary accelerated-lifetime passive soak testing has also been done in hot saline, which shows positive results for bioimplant packaging using Parylene. In addition, a neural stimulator, which features a single channel stimulation capability, has

121 been designed and the first prototype of the integrated system has been fabricated and tested to demonstrate its stimulation function in vitro.

122 APPENDIX A: DUAL-LAYER RF MEMS COIL MICROFABRICATION PROCESS

Substrate: Prime silicon wafers, 550 μm thick, 4 inch diameter. Do not remove native oxide. I. Alignment mark (Single side): 1) Hexamethyldisilazane (HMDS) primer: a. 2mL b. Time: 1minute 2) Spin AZ 4400: a. Ramp rate: 500 rpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 4) Expose in UV cannon: a. Time: 20 seconds 5) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 1 minute 30 seconds c. Rinse with DI water d. Spin dry e. Examine 6) Etch alignment marks in PEII: a. Plasma type: SF6 b. Plasma pressure/power/time: 200 mTorr/400 W/6 min (rotate wafer every 3 minutes) II. Optional sacrificial photoresist: 1) HMDS primer: a. 2 mL b. Time: 1 minute

123 2) Spin AZ 4400: a. Ramp rate: 500 rpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Hard bake: a. Temperature: 100 °C b. Time: over night III. Deposit Parylene (~ 5 μm): 1) Parylene machine I 2) Parylene C dimer: 7.5 grams IV. Optional surface treatment of Parylene: 1) PEII plasma roughening: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min 2) Surface cleaning: a. Buffered hydrofluoric acid (BHF) cleaning: 20 seconds b. Rinse with DI water c. Spin dry V. E-beam metal: 1) Material type: Ti/Au/Ti (Ti optional) 2) Metal thickness: 200 Å /2000 Å /200 Å VI. Pattern metal: 1) Spin AZ 1518: a. Ramp rate: 500 rpm/sec b. Spin speed: 3 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose in stepper: a. Time: 0.35 seconds

124 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 2 minutes c. Rinse with DI water d. Spin dry e. Examine 5) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min 6) Wet etching: a. Ti etchant: BHF ~ 20 seconds b. Au etchant: ~ 80 seconds c. Rinse with DI water d. Spin dry e. Examine 7) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with isopropanol (IPA) and DI water d. Spin dry 8) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/ 1 min VII. Deposit Parylene (insulation layer between two metal layers, ~ 2 μm): 1) Parylene machine I 2) Parylene C dimer: 3 grams VIII. Etch through vias: 1) Spin AZ 4620: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds

125 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose through via in stepper: a. Time: 6 seconds 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 4 minutes 5) Etch Parylene C in RIE: a. Plasma type: O2 b. Plasma pressure/power/time: 350 mTorr/400 W/8 min (Recipe 3) 6) Strip photoresist in ST-22: a. Temperature: 80 °C b. Time: 5 minutes c. Rinse with post-solvent rinse and DI water d. Spin dry e. Examine 7) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min IX. E-beam metal: 1) Material type: Ti /Au /Ti (Ti optional) 3) Metal thickness: 200 Å /2000 Å /200 Å X. Pattern metal: 1) Spin AZ 4400: a.

Ramp rate: 500 rpm/sec

b. Spin speed: 2 krpm c.

Time: 40 seconds

2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes

126 3) Expose in stepper: a. Time: 1.5 seconds 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 4 minutes c. Rinse with DI water d. Spin dry e. Examine 5) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min 6) Wet etching: a. Ti etchant: BHF ~ 20 seconds b. Au etchant: ~ 80 seconds c. Rinse with DI water d. Spin dry e. Examine 7) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry 8) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min XI. Deposit Parylene (~ 5 μm): 1) Parylene machine I 2) Parylene C dimer: 7.5 grams XII. Etch Parylene: 1) Spin AZ 9260: a. Ramp rate: 1 krpm/sec

127 b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose in contact aligner: a. Time: 200 seconds 4) Develop in AZ 351 developer: a. AZ 351:H2O = 1:3 b. Time: 4 ~ 5 minutes c. Rinse with DI water d. Spin dry e. Examine XIII. Etch Parylene in RIE: 1) Plasma type: O2 2) Plasma pressure/power/time: 350 mTorr/400 W/40 min (Recipe 3) 3) Examine XIV. Release devices: 1) Strip photoresist in acetone 2) Rinse with IPA and DI water 3) Release devices in DI water 4) Oven dry

128 APPENDIX B: FOLD-AND-BOND RF MEMS COIL MICROFABRICATION PROCESS

Substrate: Prime silicon wafers, 550 μm thick, 4 inch diameter. Do not remove native oxide. I. Alignment mark (Single side): 1) HMDS primer: a. 2mL b. Time: 1minute 2) Spin AZ 4400: a. Ramp rate: 500 rpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 4) Expose in UV cannon: a. Time: 20 seconds 5) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 1 minute 30 seconds c. Rinse with DI water d. Spin dry e. Examine 6) Etch alignment marks in PEII: a. Plasma type: SF6 b. Plasma pressure/power/time: 200 mTorr/400 W/6 min (rotate wafer every 3 minutes) II. Optional sacrificial photoresist: 1) HMDS primer: a. 2 mL b. Time: 1 minute

129 2) Spin AZ 4400: a. Ramp rate: 500 rpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Hard bake: a. Temperature: 100 °C b. Time: over night III. Deposit Parylene (~ 5 μm): 1) Parylene machine I 2) Parylene C dimer: 7.5 grams IV. E-beam metal: 1) Material type: Ti/Au/Ti (Ti optional) 2) Metal thickness: 200 Å /2000 Å /200 Å V. Pattern metal: 1) Spin AZ 1518: a. Ramp rate: 500 rpm/sec b. Spin speed: 3 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose in contact aligner: a. Time: 17 seconds 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 1 minutes c. Rinse with DI water d. Spin dry e. Examine 5) PEII descum: a. Plasma type: O2

130 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min 6) Wet etching: a. Ti etchant: BHF ~ 20 seconds b. Au etchant: ~ 80 seconds c. Rinse with DI water d. Spin dry e. Examine 7) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry 8) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min VI. Deposit Parylene ( ~ 5 μm): 1) Parylene machine I 2) Parylene C dimer: 7.5 grams VII. Etch Parylene: 1) Spin AZ 9260: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minute 3) Expose in contact aligner: a. Time: 200 seconds 4) Develop in AZ 351 developer: a. AZ351:H2O = 1:3 b. Time: 4 ~ 5min

131 c. Rinse with DI water d. Spin dry e. Examine VIII. Etch Parylene in RIE: 1) Plasma type: O2 2) Plasma pressure/power/time: 350 mTorr/400 W/30 min (Recipe 3) 3) Examine IX. Release devices: 1) Strip photoresist in acetone 2) Rinse with IPA and DI water 3) Release devices in DI water X. Thermal bonding and annealing: 1) Align and clamp in the appropriate jig 2) Place in the vacuum oven: a. N2 back fill b. Temperature: 220 °C ~ 250 °C c. Time: 2 days 3) Vent 4) Examine

132 APPENDIX C: EMBEDDED CHIP INTEGRATION MICROFABRICATION PROCESS

Substrate: Double-side polished silicon wafers, 550 μm thick, 4 inch diameter. Do not remove native oxide. I. Double side alignment marks: 1) HMDS primer: a. 2 mL b. Time: 1 minute 2) Spin AZ 4400 on front side: a. Ramp rate: 1 krpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Soft bake: a. Temperature: 100 °C b. Time: 10 minutes 4) Spin AZ 4400 on back side: a. Ramp rate: 1 krpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 5) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 6) Expose in UV cannon: a. Time: 20 seconds on each side 7) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 90 seconds c. Rinse in DI H2O d. Spin dry e. Examine

133 8) Etch alignment marks in RIE: a. Plasma type: SF6/O2 b. Plasma Pressure/Power/Time: 120 mTorr/120 W/10 min (Recipe 4) 9) Strip photoresist in ST-22: a. Temperature: 80 °C b. Time: ~ 10 minutes c. Rinse in post-solvent rinse and DI water d. Spin dry e. Examine II. Etch Parylene anchors: 1) HMDS primer: a. 2 mL b. Time: 1 minute 2) Spin AZ 4400: a. Ramp rate: 500 rpm/sec b. Spin speed: 4 krpm c. Time: 40 seconds 3) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 4) Expose in stepper: a. Time: 1.5 seconds 5) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 90 seconds c. Rinse with DI water d. Spin dry e. Examine 6) PEII Descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min

134 7) BHF cleaning: a. 20 seconds b. Rinse with DI water c. Spin dry 8) Etch anchor in DRIE 9) Examine 10) Do not remove photoresist: a. Hard bake: over night III. Deposit Parylene (~5 μm): 1) Protect back side with black tapes 2) Parylene machine I 3) Parylene C dimer: 7.5 grams IV. E-beam metal: 1) Material type: Au 2) Metal thickness: 5000 Å V. Pattern metal: 1) Spin AZ 1518: a. Ramp rate: 500 rpm/sec b. Spin speed: 3 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 4) Expose in contact aligner: a. Time: 17 seconds 5) Develop in AZ 351 developer: a. AZ 351: DI water = 1:4 b. Time: 1 minutes c. Rinse with DI water d. Spin dry e. Examine

135 6) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min 7) Au etching: a. Time: 150 seconds b. Rinse with DI water c. Spin dry d. Examine 8) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine 9) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min VI. Etch chip cavities: 1) Spin AZ 9260: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds, 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose in contact aligner: a. Time: 200 seconds 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes c. Rinse with DI water

136 a. Spin dry b. Examine 5) Etch Parylene in RIE: a. Plasma type: O2 b. Plasma pressure/power/time: 350 mTorr/400 W/15 min (Recipe 3) c. Examine 6) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine 7) Spin AZ 9260: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 8) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 9) Expose in contact aligner: a. Time: 200 seconds 10) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes c. Rinse with DI water a. Spin dry b. Examine 11) Remove photoresist edge 12) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min

137 13) Hard bake: a. Time: over night 14) BHF cleaning: a. Time: 20 seconds b. Rinse with DI water c. Spin dry 15) Etch chip cavity in DRIE: a. Bosch process: B=7 seconds b. 300 loops c. Examine, and measured depth in WYKO 16) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine 17) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min VII. Drop chips into cavities: 1) drop 5-minute epoxy into cavities 2) Align chips in cavities 3) Flatten wafer surface 4) Cure epoxy at room temperature VIII. Deposit Parylene (~2 μm): 1) Protect back side with black tapes 2) Parylene machine I 3) Parylene C dimer: 3 grams IX. Etch through vias: 1) Spin AZ 9260: a. Ramp rate: 1 krpm/sec

138 b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose through via in contact aligner: a. Time: 200 seconds 4) Develop in AZ 351 developer a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes 5) Etch Parylene C in RIE a. Plasma type: O2 b. Plasma pressure/power/time: 350 mTorr/400 W/8 min (Recipe 3) 6) Strip photoresist a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine 7) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min X. E-beam metal: 1) Material type: Au 2) Metal thickness: 5000 Å XI. Pattern metal: 1) Spin AZ 9620: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake:

139 a. Temperature: 100 °C b. Time: 30 minutes 10) Expose in contact aligner: a. Time: 200 seconds 11) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes c. Rinse with DI water d. Spin dry e. Examine 12) PEII descum: c. Plasma type: O2 d. Plasma pressure/power/time: 200mTorr/50W/1min 13) Au etching: a. Time: 150 seconds b. Rinse with DI water c. Spin dry d. Examine 14) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine 15) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200mTorr/50W/1min XII. Deposit Parylene (~5um): 1) Protect back side with black tapes 2) Parylene machine I 3) Parylene C dimer: 7.5 grams

140 XIII. Etch Parylene: 1) Spin AZ 9260: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 3) Expose in contact aligner: a. Time: 200 seconds 4) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes c. Rinse in DI water d. Examine 5) Etch Parylene C in RIE: a. Plasma type: O2 b. Plasma pressure/power/time: 350 mTorr/400 W/40 min (Recipe 3) 6) Strip photoresist: a. Solvent: acetone b. Time: 5 minutes c. Rinse with IPA and DI water d. Spin dry e. Examine XIV. Etch through Si from the back side: 1) Spin AZ 9260 on the front side: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 2) Soft bake: a. Temperature: 100 °C

141 b. Time: 30 minutes 3) Spin AZ 9260 on the back side: a. Ramp rate: 1 krpm/sec b. Spin speed: 1 krpm c. Time: 40 seconds 4) Soft bake: a. Temperature: 100 °C b. Time: 30 minutes 5) Expose in contact aligner: a. Time: 200 seconds 6) Develop in AZ 351 developer: a. AZ 351: DI water = 1:3 b. Time: 4~5 minutes c. Rinse in DI water d. Spin dry e. Examine 7) Remove photoresist edge on the back side 8) PEII descum: a. Plasma type: O2 b. Plasma pressure/power/time: 200 mTorr/50 W/1 min 9) Spin AZ 4400 on backing wafer: a. Ramp rate: 500 rpm/sec b. Spin speed: 2 krpm c. Time: 40 seconds 10) Stack the processed wafer on backing wafer 11) Hard bake: a. Temperature: 100 °C b. Time: over night 12) BHF cleaning: a. Time: 20 seconds 13) Etch Si in DRIE:

142 a. Bosch process: B = 7 seconds b. 760 loops c. Examine XV. Release device in acetone: 1) Time: over night 2) Rinse with IPA and DI water 3) Oven dry

143 BIBLIOGRAPHY

[1]

World Health Organization, "Magnitude and causes of visual impairment," http://www.who.int/medicacentre/factsheets/fs282/en/. (accessed: Aug. 26, 2008)

[2]

The Eye Diseases Prevalence Research Group, "Prevalence of age-related macular degeneration in the United States," Arch Ophthalmol, vol. 122, pp. 562-572, 2004.

[3]

Artificial Retina. Project, "Retinal diseases: Age-Related Macular Degeneration and Retinitis Pigmentosa," http://artificialretina.energy.gov/diseases.shtml. (accessed: Aug. 26,2008)

[4]

Wikipedia, "Retinitis pigmentosa." pigmentosa. (accessed: Aug. 26, 2008)

[5]

W.R. Green and C. Enger, "Age-related macular degeneration histopathologic studies," The 1992 Lorenz E. Zimmerman Lecture. Ophthalmology, vol. 100, pp. 1519-1535, 1993.

[6]

F.H. Verhoeff, "Microscopic observations in a case of retinitis pigmentosa," Arch. Ophthalmol., vol. 5, pp. 392-407, 1931.

[7]

E.L. Berson, "Retinitis pigmentosa," Invest. Ophthalmol. Vis. Sci., vol. 34, pp. 1659-1676, 1993.

[8]

C.A. Curcio, N.E. Medeiros, and C.L. Millican, "Photoreceptor loss in age-related macular degeneration," Invest. Ophthalmol. Vis. Sci., vol. 37, pp. 1236-1249, 1996.

[9]

K.W. Horch and G.S. Dhillon, Neuroprosthetics Theory and Practice (Series on Bioengineering & Biomedical Engineering-Vol.2): World Scientific, 2004.

[10]

E.W.D. Norton, M.F. Marmor, D.D. Clowes, J.W. Gamel, C.C. Barr, A.R. Fielder, J. Marshall, E.L. Berson, B. Rosner, M.A. Sandberg, K.C. Hayes, B.W. Nicholson, C. Weigel-DiFranco, W. Willett, J.S. Felix, and A.M. Laties, "A randomized trial of vitamin A and vitamin E supplementation for retinitis pigmentosa," Arch. Ophthalmol., vol. 11, pp. 1460-1466, 1993.

http://en.wikipedia.org/wiki/Retinitis

144 [11]

J. Bennett, T. Tanabe, D. Sun, Y. Zeng, H. Kjeldbye, P. Gouras, and A.M. Maguire, "Photoreceptor cell rescue in retinal degeneration (rd) mice by in vivo dene therapy," Nature Medicine, vol. 2, pp. 649-654, 1996.

[12]

V. Tropepe, B.L.K. Coles, B.J. Chiasson, D.J. Horsford, A.J. Elia, R.R. Mclnnes, and D.V.D. Kooy, "Retinal stem cells in the adult mammalian eye," Science, vol. 287, pp. 2032-2036, 2000.

[13]

R.E. MacLaren, R.A. Pearson, A. MacNeil, R.H. Douglas, T.E. Salt, M. Akimoto, A. Swaroop, J.C. Sowden, and R.R. Ali, "Retinal repair by transplantation of photoreceptor precursors," Nature, vol. 444, pp. 203-207, 2006.

[14]

G.S. Brindley and W.S. Lewin, "The senstaion produced by electrical stimulation of the visual cortex," J. Physiol., vol. 196, pp. 479-493, 1968.

[15]

W.H. Dobelle and M.G. Mladejovsky, "Phosphenes produced by electrical stimulation of human occipital cortex, and their application to the development of a prosthesis for the blind.," J. Physiol., vol. 243, pp. 553-576, 1974.

[16]

W.H. Dobelle, "Artificial vision for the blind by connecting a television camera to the brain," ASAIO J., vol. 46, 2000.

[17]

R.A. Normann, E.M. Maynard, P.J. Rousche, and D.J. Warren, "A neural interface for a cortical vision prosthesis," Vision Res., vol. 39, pp. 2577-2587, 1999.

[18]

D.R. Kipke, R.J. Vetter, J.C. Williams, and J.F. Hetke, "Silicon-Substrate Intracortical Microelectrode Arrays for Long-Term Recording of Neuronal Spike Activity in Cerebral Cortex," IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 11, pp. 151-155, 2003.

[19]

R. Huang, C. Pang, Y.C. Tai, J. Emken, C. Ustun, R. Andersen, and J. Burdick, "Integrated parylene-cabled silicon probes for neural prosthetics," in Proc. MEMS 2008, Tucson, Arizona, 2008.

[20]

C. Veraart, C. Raftopoulos, J.T. Mortimer, J. Delbeke, D. Pins, G. Michaux, A. Vanlierde, S. Parrini, and M.C. Wanet-Defalque, "Visual sensations produced by optic nerve stimulation using an implanted self-sizing spiral cuff electrode," Brain Research, vol. 813, pp. 181-186, 1998.

145 [21]

T. Yagi, Y. Ito, H. Kanda, S. Tanaka, M. Watanabe, and Y. Uchikawa, "Hybrid retinal implant: fusion of engineering and neuroscience," Proceedings IEEE international Joint Conference on Neural Networks, vol. 4, pp. 382-385, 1999.

[22]

M.S. Humayun, E.d.J. Jr, J.D. Weiland, G. Dagnelie, S. Katona, R. Greenberg, and S. Suzuki, "Pattern electrical stimulation of the human retina," Vision Research, vol. 39, pp. 2569-2576, 1999.

[23]

S.Y. Kim, S. Sadda, J. Pearlman, M.S. Humayun, E.d. Jr, and W.R. Green, "Morphometric analysis of the macula in eyes with disciform age-realted macular degeneration," The Association for Research in Vision and Ophthalmology annual meeting, vol. 42, 2001.

[24]

E. Zrenner, A. Stett, S. Weiss, R.B. Aramant, E. Guenther, K. Kohler, K.-D. Miliczek, M.J. Seiler, and H. Haemmerle, "Can subretinal microphotodiodes successfully replace degenerated photoreceptors?," Vision Research, vol. 39, pp. 2555-2567, 1999.

[25]

A.Y. Chow, V.Y. Chow, K.H. Packo, J.S. Pollack, G.A. Peyman, and R. Schuchard, "The artificial silicon retinamicrochip for the treatment of vision loss from retinitis pigmentosa," Arch Ophthalmol, vol. 122, pp. 460-469, 2006.

[26]

M.S. Humayun, R. Propst, E.d. Jr, K. McCormick, and D. Hickingbotham, "Bipolar surface electrical stimulation of the vertebrate retina," Arch Ophthalmol, vol. 112, pp. 110-116, 1994.

[27]

M. Javaheri, D.S. Hahn, R.R. Lakhanpal, J.D. Weiland, and M.S. Humayun, "Retinal prostheses for the blind " Ann. Acad. Med., vol. 35, pp. 137-144, 2006.

[28]

J.D. Weiland and M.S. Humayun, "Visual prosthesis," Proceedings of the IEEE, vol. 96, pp. 1076-1084, 2008.

[29]

Y.C. Tai, L.S. Fan, and R.S. Muller, "IC-processed micro-motor: design, technolog, and testing," in Proc. IEEE Int. Conf. on Micro Electro Mechanical Systems, 1989, pp. 1-6.

[30]

K.A. Shaw, Z.L. Zhang, and N.C. Macdonald, "SCREAM-i-a single mask, singlecrystal silicon, reactive ion etching process for microelectromechanical structures," Sensor. Actuat. A-Phys., vol. 40, pp. 63-70, 1994.

146 [31]

T. Stieglitz and J.-U. Meyer, Microtechnical interfaces to neurons, vol. 194: Springer, 1998.

[32]

A. Hierlemann and H. Baltes, "CMOS-based chemical microsensors," Analyst, vol. 128, pp. 15-28, 2003.

[33]

Specialty Coating Systems, "Parylene knowledge: discovery / history," http://www.scscoatings.com/parylene knowledge/history.aspx. (accessed: Sep. 2,2008)

[34]

Specialty Coating Systems, "Parylene knowledge: parylene deposition process," http://www.scscoatings.com/parylene knowledge/deposition.aspx. (accessed: Sep. 2,2008).

[35]

D.C. Rodger, "Development of flexible Parylene-based microtechnologies for retinal and spinal cord stimulation and recording," in Ph. D Thesis, California Institute of Technology Pasadena, CA 91125, 2007.

[36]

J.J. Licari and L.A. Hughes, Handbook of polymer coating for electronics: chemistry, technology, and applications, , 2nd ed. Park Ridge, N. J., U.S.A.: Noyes Publications, 1990.

[37]

Y.C. Tai, "Parylene MEMS: material, technology and applications," Proceedings of the 20th Sensor Symposium, pp. 1-8, 2003.

[38]

K.D. Wise, "Silicon microsystems for neuroscience and neural prostheses," IEEE Engineering in Medicine and Biology Magazine, pp. 23-29, 2005.

[39]

C. Pang, "Parylene technology for neural probes applications," in Ph. D Thesis, California Institute of Technology Pasadena, CA 91125, 2007.

[40]

P.W. Barth, S.L. Bernard, and J.B. Angell, "Flexible circuit and sensor arrays fabricated monolithic silicon technology," IEEE Trans. Electron. Devices, vol. ED32, pp. 1202-1205, 1985.

[41]

F. Jiang, Y.C. Tai, K. Walsh, T. Tsao, G.B. Lee, and C.M. Ho, "A fexible MEMS technology and its fist application to shearstress sensor skin," Proc. IEEE Int. Conf. on Micro Electro Mechanical Systems, pp. 465-470, 1997.

[42]

Y. Xu, Y.C. Tai, A. Huang, and C.M. Ho, "IC-Integrated flexible shear-stress sensor skin," J. MicroElectroMech. S., vol. 12, pp. 740-747, 2003.

147 [43]

Y.C. Tai, F. Jiang, Y. Xu, M. Liger, S. Ho, and C.M. Ho, "Flexible MEMS skins: technologies and applications," in Proceedings Pacific Rim MEMS Workshop, 2002.

[44]

S.C. Bayliss, L.D. Buckberry, I. Fletcher, and M.J. Tobin, "The culture of neurons on silicon," Sensor. Actuat. A-Phys., vol. 74, pp. 139-142, 1999.

[45]

S.A. Dayeh, D.P. Butler, and Z.C. Butler, "Micromachined infrared bolometers on flexible polyimide substrates," Sensor. Actuat. A-Phys., vol. 118, pp. 49-56, 2005.

[46]

J. Engel, H. Chen, and C. Liu, "Development of polyimide flexible tactile sensor skin," J. Micromech. Microeng., vol. 13, pp. 359-366, 2003.

[47]

U.K. Mudali, T.M. Sridhar, and B. Raj, "Corrosion of bio implants," Sadhana, vol. 28, pp. 601-637, 2003.

[48]

T. Stieglitz, "Methods to determine the stability of polymer encapsulations," in 10th Annual Conference of the international FES Society Montreal, Canada, 2005.

[49]

M. Kazemi, E. Basham, M. Sivaprakasam, G. Wang, D.C. Rodger, J.D. Weiland, Y.C. Tai, W. Liu, and M. Humayun, "A test microchip for evaluation of hermetic packaging technology for biomedical prosthetic implants," in Proc. IEEE-EMBS 2004, . San Francisco, CA, 2004.

[50]

A. Tooker, "Development of biocompatible Parylene neurocages for action potential stimulation and recording," in Ph. D Thesis, California Institute of Technology Pasadena, CA 91125, 2007.

[51]

T. Harder, T.J. Yao, Q. He, C. Y. Shih, and Y.C. Tai, "Residual stress in thin-film parylene C," Proc. IEEE Int. Conf. on Micro Electro Mechanical Systems, pp. 435438, Jua. 20-24 2002.

[52]

G. Bitko, D.J. Monk, H.S. Toh, and J. Wertz, "Annealing thin film parylene coating for media compatible pressure sensors," Motorola Technical developments, pp. 9294, 1996.

[53]

W.H. Hubbel, H. Brandt, and Z.A. Munir, "Transient and steady state water vapor permeation through polymer films," J. Polym. Sci., vol. 13, pp. 493-507, 1975.

148 [54]

G.E. Loeb, M. Salcman, and E.M. Schmidt, "Parylene as a chronically stable, reproducible microelectrode insulator," IEEE Transactions on Biomedical Engineering, vol. BME-24, pp. 121-129, March 1977.

[55]

Special Coating Systems, "Coatspec.pdf, http://www.scscookson.com/downloads/coatspec.pdf. (accessed: Aug. 26, 2008)

[56]

P.R. Menon, "Rates and activation energy of water vapor permeation through thin film Parylene C," in Senior Thesis, California Institute of Technology Pasadena, CA 91125, 2008.

[57]

Y. Hu, V. Topolkaraev, A. Hiltner, and E. Bear, "Measurement of water vapor transmission rate in highly permeable film," Journal of Applied Polymer Science, vol. 81, pp. 1624-1633, 2000.

[58]

D.R. Lide, Handbook of chemistry and physics: CRC Press, 1996.

[59]

P. Singh and P. Viswanadham, Failure modes and mechanisms in electronic packages: Springer, 1997.

[60]

A. Tanioka, N. Fukushima, K. Hasegawa, K. Miyasaka, and N. Takahashi, "Permeation of gases across the poly (chloro-p-xylylene) membrane," Journal of Applied Polymer Science, vol. 54, pp. 219-229, 1994.

[61]

T.P. Hoar, "Review lecture: Corrosion of metals: Its cost and control " Proc. R. Soc. Lond. A. Mathermatical and Physical Sciences, vol. 348, pp. 1-18, 1976.

[62]

D.M. Dragutin and J.P. Popic, "Electrochemical and chemical corrosion of chromium," B. Chem. Tech. MKD, vol. 23, pp. 93-100, 2004.

[63]

J.D. Weiland, W. Liu, and M.S. Humayun, "Retinal Prosthesis," Annu. Rev. Biomed. Eng., vol. 7, pp. 361-401, 2005.

[64]

G. Wang, W. Liu, M. Sivaprakasam, and G.A. Kendir, "Design and analysis of an adaptive transcutaneous power telemety for biomedical implants," IEEE Transactions on Circuits and Systems-I: Regular Papers, vol. 25, pp. 2109-2117, 2000.

[65]

C. Kim and K. Wise, "A 64-site multishank CMOS low-profile neural stimulating probe," IEEE Journal of Solid-State Circuits, vol. 31, pp. 1230-1238, 1996.

"

149 [66]

B. Ziaie, M. Nardin, A. Coghlan, and K. Najafi, "A single-channel implantable microstimulator for functional neuromuscular stimulation," IEEE Transactions on Biomedical Engineering, vol. 44, pp. 909-920, 1997.

[67]

G.A. Kendir, W. Liu, G. Wang, M. Sivaprakasam, R. Bashirullah, M.S. Humayun, and J.D. Weiland, "An optimal design methodology for inductive power link with class-e amplifier," IEEE Transactions on Circuits and Systems-I: Regular Papers, vol. 52, pp. 857-866, 2005.

[68]

Z. Hamici, R. Itti, and J. Champier, "A high-efficiency power and data transmission system for biomedical implanted electronic devices," Meas. Sci. Technol. , vol. 7, pp. 192-201, 1996.

[69]

J.D. Weiland, Personal Conversation.

[70]

Health Canada, "Guidelines for limiting radiofrequency exposure-short wave diathermy-safety code 25," in Biological effects of RF exposure, http://www.hcsc.gc.ca/ewh-semt/pubs/radiation/83ehd-dhm98/effects-biolog-effets-eng.php. (accessed: Sep. 4, 2008)

[71]

H.B. Dwight, Electrical coils and conductors: McGraw-Hill Book Company, INC. , 1945.

[72]

D.K. Cheng, Field and wave electromagnetics: Addison-Wesley, 1989.

[73]

P.J. Dowell, "Effects of eddy currents in transfromer windings," Proc. Inst. Elect. Eng., vol. 133, pp. 1387-1394, 1966.

[74]

C. Wu, C. Tang, and S. Liu, "Analysis of on-chip spiral inductors using the distributed capacitance model," IEEE Journal of Solid-State Circuits, vol. 38, pp. 1040-1044, 2003.

[75]

A. Zolfaghari, A. Chan, and B. Razavi, "Stacked inductors and transformers in CMOS technology," IEEE J. Solid-St. Circ., vol. 36, pp. 620-628, 2001.

[76]

O. Kenneth, "Estimation methods for quality factors of inductors fabricated in silicon integrated circuit process technologies," IEEE J. Solid-St. Circ., vol. 33, pp. 1249-1252, 1998.

[77]

W.H. Ko, S.P. Liang, and C.D.F. Fung, "Design of radio-frequency powered coils for implant instruments," Med. Bio. Eng. Comput., vol. 15, pp. 634-640, 1977.

150 [78]

J. Wu, V. Quinn, and G.H. Bernstein, "A simple, wireless powering scheme for MEMS devices," Proc. SPIE, vol. 4559, pp. 43-52, 2001.

[79]

F.E. Terman, Radio engineers' handbook. New York: McGraw-Hill, 1943.

[80]

"Feasibility study of power telemetr using MEMS coil for intraocular retinal prosthesis," University of California, Santa Cruz.

[81]

E. Meng and Y.C. Tai, "Parylene etching techniques for microfluidics and BioMEMS," in Proc. IEEE int. Conf. on MEMS, 2005, pp. 568-571.

[82]

D.C. Rodger, A.J. Fong, W. Li, H. Ameri, I. Lavrov, H. Zhong, S. Saati, P. Menon, E. Meng, J.W. Burdick, R.R. Roy, V.R. Edgerton, J.D. Weiland, M.S. Humayun, and Y.C. Tai, "High-density flexible parylene-based multielectrode arrays for retinal and spinal cord stimulation," in Solid-State Sensors, Actuators and Microsystems Conference 2007, 2007, pp. 1385-1388.

[83]

P.J. Chen, W.C. Kuo, W. Li, Y.J. Yang, and Y.C. Tai, "Q-enhanced fold-and-bond MEMS inductors," Proc. IEEE Int. Conf. on Nano/Micro Engineered and Molecular Systems, pp. 869-872, 2008.

[84]

J.N. Burghartz, "Integrated multilayer RF passives in silicon technology," Silicon Montolithic Integrated Circuits in RF systems, 1998. Digest of Papers. 1998 Topical Meeting, pp. 141-147, 1998.

[85]

J.-U. Meyer, T. Stieglitz, O. Scholz, W. Haberer, and H. Beutel, "High Density Interconnects and Flexible Hybrid Assemblies for Active Biomedical Implants," IEEE Transactions on Advanced Packaging, vol. 24, pp. 366-374, 2001.

[86]

D.C. Rodger, J.D. Weiland, M.S. Humayun, and Y.C. Tai, "Scalable high leadcount parylene package for retinal prostheses," Sensor. Actuat. B-Chem., vol. 117, pp. 107-114, 2006.

[87]

M. Liger, D.C. Rodger, and Y.C. Tai, "Robust parylene-to-silicon mechanical anchoring," in Proc. 16th IEEE Int. Conf. on Micro Electro Mechanical Systems, Kyoto, Japan, 2003, pp. 602-605.

[88]

N.A. Sachs and G.E. Loeb, "Development of a BIONic muscle spindle for prosthetic proprioception," IEEE T. Bio-med. Eng., vol. 54, pp. 1031-1041, 2007.

151 [89]

A.C. Dupont, S.D. Bagg, J.L. Creasy, C. Romano, D. Romano, F.J.R. Richmond, and G.E. Loeb, "First clinical experience with BION implants for therapeutic electrical stimulation," Neuromodulation, vol. Vol. 7, pp. 38-47, 2004.